There are many ways to play Pick 3 and even more ways to track and analyze the numbers.I don’t play boxed much, but when I do, I often refer to the following information.This info is all based on probability that is reinforced by statistical performance.Much of this data is probably common knowledge for the more experienced players, but for many others it may provide a whole new approach to picking boxed combinations.

This method turns the art of anticipating boxed combinations into a strategy that is similar to analyzing the straights.When looking at the previous results to choose straight numbers, many players will use positional analysis to pick certain digits that will hit in specific positions.This is often accomplished through frequency analysis of the digits, skip averages, max-out or other such factors like the anticipation of a repeating digit, etc.Many times, the front, back and split pairs are often considered when deciding straight picks as well.

Its seems to me that much of this deciding analytical criteria is often left out of boxed tracking mainly because the digit order or their positions are not so easily observed.More often than not, order is not observed during boxed play.After all, the order of the digits doesn’t really matter when deciding if your boxed/any-order wager is a winner.However, this does not mean that the concept of order is unimportant when deciding what boxed numbers to play!

For this strategy, the concept of order is not principally based on position one, two and three…as is done during straight play, but rather on the three key factors of Low, Medium and High.Every boxed number has a lowest digit drawn, a highest digit drawn, and a medium digit drawn.The medium digit usually hangs somewhere in between the low and high digits, but sometimes it is equal to one of them.Arranging all the winning straights/exacts in lowest to highest format, from left to right, can create an observable and trackable positional order that can effectively be used for boxed play.Arranging straight/exact numbers in this format is easy:

EXACTBOXED 1 2 3L M H 3-0-20-2-3 7-8-97-8-9 1-9-41-4-9 3-2-22-2-3 8-9-88-8-9 5-5-55-5-5 9-0-90-9-9

If the exact number that is drawn is already arranged from lowest to highest order, then it stays the same and there’s no need to arrange it (like 789 above).If the number is a Triple it also stays the same.If the number is a Double, put the two lowest digits to the left or the highest two digits to the right.In the list above, the bold 1 2 3 under “EXACT” represent the positions one, two and three as they were drawn straight.The bold L M H under “BOXED” indicates the Lowest, Medium, and Highest digits drawn after being rearranged.In the case of a Double-Digit combination, if the duplicated digits are higher than the non-duplicated digit, they go to the right…if they are lower, then they go to the left.If you keep you’re historical listing of straight winners on a Microsoft Excel file, you can easily have your computer create a low/medium/high listing for your entire list in a matter of seconds.I won’t get into how to do this unless someone actually posts a reply needing to know how to do so.

Once you have your historical listing of the straights arranged for boxed analysis in the L/M/H format, you will immediately notice the overabundance of 0’s and 1’s in the LOW column and a proportionately high surplus of 8’s and 9’s in the HIGH column.If you glance through the MEDIUM column, you will notice how the digits tend to appear to be more evenly distributed…and even though this distribution is a little more uniform, under closer scrutiny you will find that the digits 4 and 5 account for about 29.6% of all digits found in the medium column.

Below is a sample of a basic listing of straight numbers with L/M/H boxed numbers to the right of them.I used Excel’s conditional formatting to bold the 0’s and 1’s in the low column and the 8’s and 9’s in the high column.

As you can see, the LOW column is loaded with 0’s while the High column is cram-full of 9’s.The 1’s and the 8’s appear in their respective columns but to a somewhat lesser extent.The 0’s and 1’s kind of seem to take turns being “streaky”…and in a parallel fashion, so does the 8’s and 9’s.If you look through your own list, you will probably notice this happening from time to time.

Rearranging the straight into this low-to-high format allows you to pick your boxed combinations based on which digit you think will be the lowest or highest digit drawn.This is where boxed positional analysis comes in.If you are going to guess that either the digits 0 or 1 will be the lowest digit drawn, try to base that guess solely on how each performed in the LOW position.You can normally disregard their performance in both the MEDIUM and HIGH positions, but not always.The same goes for the 8 and 9 in the HIGH position, you can normally ignore their performance in the medium and low positions…but only most of the time.

Unlike straight positional analysis, where each digit has equal probability in each column, boxed positional tracking required you to pay close attention and understand the real probabilities at work—because the various digits have different probabilities of appearing within the three columns!This is obviously why the LOW column is loaded with 0’s and 1’s and the High column with 8’s ad 9’s.Once you understand the probabilities, it can make getting one digit correct a cakewalk and two digits moderately routine…and all this while only playing only a few combinations per game.Now, I am not going to claim that this strategy is the easiest or best in the world, but I do believe that if you study it and use it often enough, that you will find yourself pinpointing boxed wins with only a few combinations quite often.

All the probabilities associated with this strategy are calculated using straights.What I mean is that even though we are tracking or guessing boxed numbers, the probability of the digits and/or pairs appearing in the specific L/M/H categories are based on the number of straights that could be drawn and then be rearranged to meet the criteria of the event.As an example, the chance or probability that the digit 2 will be drawn as the lowest number in a boxed combination is 16.9%.This percentage is based on the fact that if the 2 were to be the lowest digit drawn, then neither the 0 nor the 1 could be drawn with it, which leaves us with 126 no-match straights, 42 double-digit straights and lastly, 1 triple-digit straight.These quantities add up to 169 straights, which is 16.9% of the total 1,000 Pick 3 numbers.

The chart below lists the chances for each digit to appear in each of the L/M/H positions.The columns labeled “AMT.” are the amounts of straights that could be drawn to make that digit appear in the specific column after the straight is rearranged into the L/M/H format.These amounts are then divided by 1,000 in order to give the probability of the digit appearing.

You will probably notice that the High and Low columns are reflections of each other.Notice that there is only 1 straight that can put a digit 9 in the low column.That number is 999.Similarly, the only straight that can cause a 0 to fall in the high column is 000.If you look at the percentage of each digit in each column, it becomes clear what you can expect to see and at what percentage of time.By adding the two percentages for 0 and 1 in the low column together, you can see that a 0 or 1 should appear in the low column 48.8% of all games.The same can be said for the 8 and 9 in the high column.In 10,859 consecutive games in Ohio the digits 0 and 1 accounted for 49.23% of all digits drawn in the low column and the digits 8 and 9 accounted for 49.07% of all digits drawn in the high column.No matter which state you are in, the percentages should closely follow the probabilities listed in this chart.

Of course, simply guessing the 0 or the 1 as the lowest digit or the 8 or 9 as the high digit isn’t going to guarantee you a win.You’ll need to try and foresee the less probable digits in both the low and high positions and also anticipate the more erratic medium digits as well.But in the event that you are guessing for a 0 or 1 in the low and/or the 8 or 9 in the high, here is one very useful observation you’ll definitely want to know about…I call it the “01 89 Repeat Stack.”

Here’s how this little observation works…The 01 Stack: when you see a 0 in the low position that is immediately followed by a 1 (in the next games low position), OR when you see a 1 in the low position followed by a 0 in the next games low position…the chance that you will see a 0 or 1 within the next three games is 86.58%.That figure is based solely upon mathematical probability, which is why it also performed right on par at 87.72% in Ohio’s 10,859 games.Then there is The 89 Stack.This works exactly like the 01 Stack but occurs with the 8 and 9 in the high position.When you witness an 8 followed by a 9 or vice versa, know that one of the two has an 86.58% chance of showing up as the high digit sometime within the next three games.In Ohio, the 89 Stack performed within expectations at 85.87%.Below is a small and very basic example of how the stacks may appear.The bolded digits represent the stacks.Stacks will often overlap each other within the same column and sometimes both the 01 and 89 stacks may occur simultaneously.

02/20/07236

02/21/07156

02/21/07009

02/22/07079

02/22/07048

02/23/07026

02/23/07269

02/24/07057

02/24/07157

02/26/07278

02/26/07024

02/27/07013

02/27/07199

Guessing the 0 or the 1 is pretty easy.The same goes with for the 8 and the 9.Given the probabilities of each of these digits to occur in their respective positions, you might be thinking that you can just simply play the pairings of 0X8, 0X9, 1X8, and 1X9 and only have to worry only about that one lousy digit that will be drawn as the medium X.Unfortunately, it’s not so simple, but you will notice that one of these four pairings will occur quite regularly.I should also add that these probabilities are mutually degenerative.To expand on what I mean by this I will give you an example.Lets suppose that you somehow had advance knowledge that the lowest digit to be drawn in the next game was going to be a 0.This knowledge can come via of the voice of God, a premonition from Nostradamus, or even Jedi-type knowledge of the force...LOL.The point is that IF the low digit is a 0, then the 9 does not truly retain its 27.1% probability to be drawn as the high digit during that same game.Why?Because with the 0 drawn as the low digit there are only 48 no-match straights and 6 doubles-digit straights (this totals 54 or 5.4%) that could be drawn to give you both the 0 as the low digit and the 9 as the high digit simultaneously.So don’t go putting both these eggs into the same basket.

As for the other digits, it will take you a little more intuition to get them right.I could go into skip averages and the like, but I won’t because I can’t say that I’ve every really found them too useful, but perhaps someone else does and can and offer their insight.To me, averages are misleading, as they are comprised of all skips, both short and long and everything else in between.With this in mind, its no wonder that only a very limited number of an event’s hits will occur close to its skip average.Without using averages, I’m often left to employ nothing but pure chance...and I like to keep that chance as close to a real equilibrium as possible.To do this, I play within the probability median.For those who aren’t familiar with the term, the median is the middle…a halfway point… a center of balance, or as its observed in the world of gambling: the 50/50 split, the 50% chance mark and finally: the probability median.Every single event in lotto can be given a median.It doesn’t matter what the odds are or what the observable event is, everything has a 50% chance!This strategy is no different.

You may find it very beneficial to try and anticipate the digits according to their specific probability median.Each of the digits have different median values in each of the different L/M/H positions.I should stress that these medians are calculated using a specific probability formula.It is important to realize that these medians are calculated first and only afterwards is an analysis of a games performance measured against them.Because the medians are, in fact, based on mathematics, you can expect every state’s Pick 3 to closely adhere to them.To say that any event wont closely follow its calculated median would be something like saying that doubles wont be drawn close to 27% of the time…so know that medians ARE accurate!

As I said before, medians are the calculated 50% chance mark for certain observations or events to happen within.Lets look at the digit 2 being drawn in the low position for example.Its actual median value is 3.74 trials.What this means is that there is a 50% chance that the digit 2 will be drawn sometime within the next 3.74 consecutive games.Obviously, we can’t play exactly 3.74 games so we must bump this number up and play the full 4.By bumping this number up to 4 games even, the chance actually goes up to a little higher than 50%.The exact probability is now 1-(1-.169)^4.This equates to about 52.31%.Realistically, you can count on the medians to be true no matter where you start to anticipate them.For example, lets say that the digit 2 has not been drawn for ten back-to-back games as the low digit.The chance that it will be selected as the low digit within the next four games is still 52.31%…even though it’s already out ten.However, as part of this strategy you should only try to anticipate a digit hitting within its median when it was the last digit drawn or at the very most within a few games ago.For example, if the low digit drawn in the LAST MOST RECENT game was a 2, you can anticipate with a 52.31% chance that it will be drawn again as the lowest digit sometime within the next four games.The idea is to play off the digits last occurrence with hopes that it hits within its median.As another example: the digit 5 in the low position is given 12 consecutive games for its median value.If the 5 occurred eight games ago, it is still viable to hope that it hits within its median, which still has an additional 5 games to go before its over.

The charts below outline the medians of each of the digits for the L/M/H positions.These charts also include some real-life data from Ohio’s 10,859 games.In the “Percent of Hits in Median” column you can see just how strongly the skips of Ohio’s digits followed the given median values.Again, the actual median values had to be rounded up to the next whole game, so most of the totals and their true probabilities are slightly above the 50% mark.

As kind of a sub-system to this strategy, you can keep track of each digits streaks in each position.This is similar to recording long series of heads or tails results when coin flipping.These streaks appear something like this: xxx--xx-x-x---xxx--xx etc.If you track the streaks you’ll gain even more insight when deciding on what digits to play.Tracking the streaks is easy.Simple record each time a digit hits within its median as an “x” and each time that it doesn’t with a “-“ (the dash mark).Below is an example using a short skip history of the digit 2 in the low position.Note: the oldest draw (23) is on the left and the most recent draw is on the right.

The median given median value for the digit 2 in the low position is 4 games.Each skip that is less than or equal to 4 is recorded with an X to indicate a win.Any skip that greater than 4 is given a “-“ to indicate a loss.The string of skips above converts to a series of wins or losses that appears as follows:

-x---xx–x–xxx--xxxxx--x-

When you start tracking these streaks you’ll essentially be anticipating a heads are tails type of outcome…but with a slight bias for heads (a win).The bias is due to the fact that the medians had to be bumped up to whole amounts of games which made the probabilities just slightly over just 50% for most of the digits.You could be making same rather lengthy charts of you do this on paper, so you may want to create a spreadsheet to do this for you instead.You’ll have to play with the strategy and get a feel for.Once you do, it will really start to become an effective tool.

Aside from the individual digits and their related positional probabilities and medians, there is also Low, Medium and High Pairs.These pairs are similarly structured to the front, back, and split pairs that are found in the straight game.When you look at your rearranged list of boxed numbers, the low and medium digits form the Low-Pair.The medium and high digits form the High-Pair.Lastly, the low and high digits form the Medium-Pair.There is a total of 55 boxed pairs, each of which are capable of being drawn in one of the three low, medium or high positions.And even though each one is capable of hitting in the three positions, certain pairs are much more likely to be drawn in certain positions than others are.The table below lists the 55 pairs and their probabilities to hit within the corresponding positions.

Examine the pairs and their percentages closely.You’ll see how the probability of each pair changes based on which pair type it is expected to be…that being the low, medium or high pair.Each of these pairs also has its own median, although I have not listed them on this table.Perhaps the best overall median for anticipating boxed pairs is the pair’s true probability median itself.Temporarily set aside the concept of low, medium and high for a few second here.What I mean by a pairs true median is its 50% chance to hit ANYWHERE within a combination, regardless of whether it’s the lowest, medium or highest pair drawn…just as long as its there. There are really only two types of boxed pairs…the no-match pairs and the double-digit pairs.The probability median for a no-match pair is 13 games (rounded up from 12.49) and the median for a double-digit pair is 25 games (rounded up from 24.41).

Lets say the last combo drawn was 023.There are three no-match pairs within that combo: 02, 03 and 23.Pick whatever one of the three pairs that you like the most, but either way, each of them by itself has a 50% chance of being drawn again sometime within the next 13 games.We may not know what position the pair will hit in, but nonetheless, the chance is still 50% over the next 13 games.

Double-digit pairs work the same way.Say 577 was the last combo drawn.There is a 50% chance that we will see double 7’s drawn again within the next 25 games.Have you ever noticed how a particular set of doubles hits and then hit again soon afterwards?This is caused by the probability median.The same goes for the no-match pairs.Most of the time, the streaks or clusters of digit pairings that you see can be attributed to the repeated success of a pair hitting within its median…kind of like seeing heads thrown three or four times in a row.

The gist of this system chiefly focuses on tracking the primary digits of 0 and 1 in the low position and the 8 and 9 in the high position.Even though you’ll benefit from tracking the less probable digits like 2 and 3 in the low or 6 and 7 in the high, you can count on at least one of the four primaries hitting 78.4% of the time.If at least one of the four primaries is hitting 78.4% of the time, then you can also bet the remaining 21.6% of the time that none of the four digits are hitting at all.

So what happens when neither the 0 nor the 1 is drawn as the lowest number and also neither the 8 or the 9 is drawn as the highest?This event (I call it the inversion) brings the total possibilities down to 6^3, which equates to only 216 straights.Since we are playing a boxed/any-order game, we can reduce those 216 straights down to 56 boxed combinations.These combinations consist of the following:

·6 Triples

·20 No-Match Combos

·30 Double-Digit Combos

Now, if you can accurately anticipate when the inversion is going to hit, you can use some other probabilities and tracking techniques to narrow your picks down even further.We all know that the chance for a double is 27% for each and every drawing…but if you correctly foresee the inversion, then there is a 41.67% chance that a double will hit when the inversion comes!In other words, if none of the four digits that we are tracking in low/high position (0-1 / 8-9) are selected in the next draw, then the chance to see a double climbs up to 41.67%.If you look at the at how the 56 combos listed above break down, that percentage may not seem quite accurate, but remember the 30 doubles represent 90 straights and the 20 no-match represent 120 straights.Since there are 216 straights that can cause the inversion to happen, then 90÷216 = 41.67%.In that same way, the 120 no-match straights have a probability of 120÷216 = 55.56%.The 6 triples are left with a 2.78% (6÷216) chance of occurring during the inversion.

If you knew the inversion was going to hit, you could play all 56 boxed numbers for a small any-order win, but we since can never really know for sure, it would probably be best to just pick and choose a much smaller amount of combos from that group based on certain criteria.This can be by positional analysis of the low/medium/high digits (other than (0-1-8-9) or even by boxed pairs or the skips of the combos themselves.Here is an interesting way to break the 56 boxed inversion combos down…and that is by consecutive and non-consecutive pairs occurring within the combos!

When the inversion hits, there is a 58.33% chance that a consecutive pair will be contained within the winning number!This chance is the combined probability of the 13.89% chance of a double-digit combo containing a consecutive pair and the 44.44% probability of a no-match combo containing a consecutive pair.If you are anticipating the inversion to hit the next drawing, I strongly recommend playing the no-match combinations that contain a consecutive pair along with the doubles that DO NOT contain a consecutive pair.Here’s why: the probability for a double that contains a consecutive pair is 13.89%, while the probability for a double that DOES NOT is 27.78%!If a double is drawn during the inversion, you’re twice as likely to NOT see it contain a consecutive pair.On the other hand, the no-match combos are completely opposite.A whopping 80% of the no-match inversion numbers contain a consecutive pair, whereas only 20% do not!So, playing this best overall mix of numbers should put you in the hit zone 72.22% of the time.

Think about it…with the omission of the digits 0,1,8, and 9 from play, there are only 5 consecutive pairs that can hit, they are: 23, 34, 45, 56, and 67.By the way, there are only four no-match combos in the inversion group that don’t contain them (246, 247, 257, and 357).

The inversion group itself has a 21.6% chance of occurring.That’s a little more than once in every five games, so it’s helpful to understand the probabilities associated with it. There are many statistics and probabilities that can be utilized to help anticipate when the inversion is going to hit.Perhaps the most useful is the events probability median, which lies at about 2.85 games.Since it is impossible to play only 2.85 games, we can bump this number up to 3, which gives us a 51.81% chance within 3 games.

I could go on and on with different ways to use this strategy.Your best bet is to create a list of previous results in the low/medium/high format and try it out for a while.Depending on the interest this post generates, I may be adding some other charts and info to help further explain and illustrate the probabilities involved with this strategy.

The Carolinas - Charlotte United States Member #21627 September 12, 2005 4138 Posts Offline

Posted: March 5, 2007, 1:24 pm - IP Logged

This is quite interesting...I would think that applying skip analysis and progressive wagering could allow some efficient profits over the long haul, whether it be online or at the local lottery retailer.

The North Carolina Education Lottery - so much a joke that here are their mascots:

This stuff is for advanced users of filters, I hope that I am not confusing people.

I am trying to make up for my rude posts above, by showing some advanced filter techniques.

The same can be done for the Boxed or Any Order Numbers' Matrix.

This stuff goes even way beyond that, but it is way too complicated for most people and gets either too hard to post or impossible, very hard to understand for most if I go beyond that (ABOVE).

This stuff is for advanced users of filters, I hope that I am not confusing people.

I am trying to make up for my rude posts above, by showing some advanced filter techniques.

The same can be done for the Boxed or Any Order Numbers' Matrix.

This stuff goes even way beyond that, but it is way too complicated for most people and gets either too hard to post or impossible, very hard to understand for most if I go beyond that (ABOVE).

By the way, on the above, the oldest draw would be down and the newest way on top.

The Carolinas - Charlotte United States Member #21627 September 12, 2005 4138 Posts Offline

Posted: March 5, 2007, 5:20 pm - IP Logged

Thoth

I have read this all the way through and have found it very intriguing. What would the median be of a 0 or 1 hitting after a 0 or 1 has hit in the low position?

The North Carolina Education Lottery - so much a joke that here are their mascots:

Findlay, Ohio United States Member #4855 May 28, 2004 400 Posts Offline

Posted: March 5, 2007, 8:21 pm - IP Logged

Quote: Originally posted by LANTERN on March 5, 2007

Sorry, I didn't mean to sound as bad as that sounds, let me know if I can help some more.

Oh it's OK, I didnt think it was rude.

I had originally thought about posting info on the combos "width" too...I call it distance. I figured if this post had enough interest I'd post this data and also the expected max skips.

Tonights Ohio draw was 480 straight...which is 048 as LMH boxed. It was probably an easy hit for any one already familiar with a similar stratyegy.

Findlay, Ohio United States Member #4855 May 28, 2004 400 Posts Offline

Posted: March 5, 2007, 8:34 pm - IP Logged

Quote: Originally posted by cps10 on March 5, 2007

Thoth

I have read this all the way through and have found it very intriguing. What would the median be of a 0 or 1 hitting after a 0 or 1 has hit in the low position?

The actual probability for a 0 OR a 1 (in the low position) is .488 (48.8%) for any single drawing. So the actual 50% mark is 1.035 games. Again, we cant just play that ending .035 of a single game, so the actual amount of games has to be left at either one, for the 48.8% chance, or bumbped up to two whole games for a 73.79% chance.

The Carolinas - Charlotte United States Member #21627 September 12, 2005 4138 Posts Offline

Posted: March 5, 2007, 8:38 pm - IP Logged

Quote: Originally posted by Thoth on March 5, 2007

The actual probability for a 0 OR a 1 (in the low position) is .488 (48.8%) for any single drawing. So the actual 50% mark is 1.035 games. Again, we cant just play that ending .035 of a single game, so the actual amount of games has to be left at either one, for the 48.8% chance, or bumbped up to two whole games for a 73.79% chance.

Interesting...so that would be 1 - ((1-0.488) * (1-0.488))?

So a good rule of thumb would be if you haven't seen the 0 or 1 in low position within 3 draws, then things might be in order to play them?

Similarly, if a 0 or 1 hits in 3 or 4 straight draws, it might be time to play the inversion?

The North Carolina Education Lottery - so much a joke that here are their mascots:

Findlay, Ohio United States Member #4855 May 28, 2004 400 Posts Offline

Posted: March 5, 2007, 9:12 pm - IP Logged

Quote: Originally posted by cps10 on March 5, 2007

Interesting...so that would be 1 - ((1-0.488) * (1-0.488))?

So a good rule of thumb would be if you haven't seen the 0 or 1 in low position within 3 draws, then things might be in order to play them?

Similarly, if a 0 or 1 hits in 3 or 4 straight draws, it might be time to play the inversion?

1-(1-0.488)^2 is correct

Yes, three or four draws without the 0 or 1 in the low is a good time to play them. Of course. when you see four of five games in a row where 0 or 1 was drawn, I'd say start looking at the other digits like 2 and 3. The same goes for 8 and 9 on the high side.

As for the inversion...it seems like it comes and goes alot. I like to wait until the inversion actually hits, then I play numbers from that group with the anticipation that the inversion will hit within its median, which is sometime within the next 3 games. Remember, 50% of the time when an inversion hits, another one is coming within the next three games. If you dont wait for that first hit, you might be better off waiting until you see a large stack of back-to-back games where the 0 or 1 was drawn in the low and 8 or 9 was drawn in the high. Large streaks of games that continually produce 0 or 1 in the low and or and 8 or 9 in the high eventually have to end...and when they do, thats when it hits.

Take a look at how many consecutive games where one of the four primary digits hits at either end. Dont look at just the low or at just the high...look at both. If you see 10 consecutive games with the primaries hitting in their positions, the inversion is coming very, very soon.

Anywhere & Everywhere United States Member #10713 January 23, 2005 290 Posts Offline

Posted: March 5, 2007, 10:05 pm - IP Logged

Quote: Originally posted by LANTERN on March 5, 2007

Just some extra info:

The Lowest-Middle-Highest filter:

LMH = BS1 (Boxed Single 1)

LHM = BS2 (Boxed 2)

MHL = BS3 (Boxed 3)

MLH = BS4 (Boxed 4)

HLM = BS5 (Boxed 5)

HML - BS6 (Boxed 6)

Because the singles are 6 way numbers

For doubles as they are 3 way numbers:

DDS = BD1 (Boxed Double 1)

DSD = BD2

SDD = BD3

D = Double Digit Pair

S = Single Digit.

Or:

LLH

HLL

LHL

HHL

LHH

HLH

H = Highest Digit(s)

L = Lowest Digit(s)

I made a change to the post

Lantern, thanks for adding a few little nuggets to this post. These are to me fundamentals which are very important. Is there more, should I send you PM on the matter for the advanced filters, maybe we can discussed in more depth offline.