Random events are supposed to be randomly distributed which means in a given period of time all events should occur the same (or at least very similar) number of times. That's the core of the definition of randomness. The time frame plays a very important role. The longer the time frame the more evenly the events will be distributed. This may not be so in short periods of time where the distribution of events may appear far from random. Frequently, observable patterns may emerge, seemingly contradicting randomness. In fact, the presence of patterns in short periods of time is THE RULE in random distribution. Over the time opposite patterns will cancel each other out giving an appearance of perfect (or near perfect) randomness.

If the patterns can be observed, can they be predicted as well? This depends on the number of events that participate in the random distribution. A coin toss has only 2 such events (heads or tails) and thus patterns are easy to observe and predict with a high degree of accuracy. A cube die which has 6 numbers (1 to 6) is a much tougher nut to crack. The point is, the greater the number of events for random distribution the more difficult it is to observe and predict patterns.

Why is this important? All numeric lotteries are random number distribution systems. As any other random system they will create patterns. And the fewer the numbers the lottery has the easier it is to observe and predict its patterns.

All Pick 3 lotteries have 2 pools of events for random distribution: 10 digits (0 to 9) and 1000 combinations (000 to 999). In which of these 2 pools will it be easier to observe and predict patterns? As the answer is no-brainer the pool of combos can safely be discarded as the lost cause by design. It always puzzled me why people so often look for patterns between combos - and then complain that these patterns are difficult to observe and short lived.

So we are stuck with the pool of 10 digits. Each Pick 3 draw selects only 3 digits from the pool. If you also select only 3 digits for playing the odds of hitting the right combination are only 1 in a 1000. Definitely not too good.

But who says I have to select only 3 digits from the 10? How about if I pick 4? Obviously, they have to be distributed into 3 digit combinations - they have to be WHEELED. How about 5 digits out of 10? The same story - you need a wheel. And so on.

How far can you go with wheeling digits? As far as your budget will allow you. The more digits you wheel the bigger the wheel. The type of play also plays a significant role as straight plays require much bigger wheels than box plays.

But why bother with wheels in the first place? Because they GUARANTEE wins. If a single was drawn (all 3 digits different), you played 5 digit single wheel and all 3 are in your 5 digits selection you have a STRAIGHT WIN no matter what's the sequence of the digits drawn. 5 digit require 60 plays $1 each. The prize may vary in different lotteries but $500 is a standard. If you distribute your 5 digits helter-skelter into the same 60 plays you may match all 3 as well - and end up winning nothing if you don't have the right combination. And that's the difference - $500 difference. If you are serious about making money in Pick 3 wheels are a MUST.

So we look for patterns among 10 digits. You can use whatever method you prefer for identifying patterns. Personally I prefer simple statistics because they are easy to compute and verify - and they work. This presentation will tell you how.

For the digit analysis I used data randomly generated by an app which used Android random engine. The process was fully automated and I did not have any influence on the randomization process. 250 draws were generated and as the screen below indicates the win types were distributed fairly randomly, very close to the expected averages.

This data file was used for all the computations and backtracks therefore comparisons between them are relevant and reliable. In this presentation backtracks are the most important as they clearly demonstrate relationships between consecutive draws.

Backtracks used certain preset values for cost and wins for computing return on investment (ROI). The screen below demonstrates these values.

Originally, what I was looking for were patterns between 10 digits depending on different criteria otherwise defined as prediction methods. What I ended up with, unexpectedly, are relationships between actual winners - within the same prediction methods. Backtracks revealed these relationships which otherwise would remain hidden.

Computerization of backtracks was essential. It would take an immense amounts of time and efford to compute these results manually, and reliably for that matter, as errors could completely distort the picture.

As mentioned earlier the more digits are wheeled the bigger the wheel becomes. We have to balance the cost of wheel with the prize it may potentially win. For straight play, either single or double, the maximum number of digits is 5, otherwise the wheels become too large and will not balance the cost. For box plays, however, the maximum number of digits is 8 (yes, that's 80% of the whole pool). And this makes a huge difference when it comes to the number of wins in each type of play.

Let's start from wheeling 5 digits for all types of plays. Below are the results for one prediction method: cold numbers selected from 15 draws, analysed for 30 draws. These results, including ROI, are typical. Still far from even breaking even.

Wins come but they are spread apart, without a pattern or consistency. Not much to get excited about.

Now let's switch to wheeling 8 digits. The same data, the same time frame as before. Repetitive digits - digits that were drawn at least twice starting from the latest draw and going back. Let's see the totals first. Below are the results.

Financial results definitely not satisfactory by any standard. But that's not what matters. Look at the number of winners for singles and doubles. They hit 19 times in 30 tries. What it means is that they had to CLUSTER, several wins in a row. And how long these clusters could be? Let's see on the screens.

 

There were 2 clusters, clearly visible: draws 3 - 15 with 13 wins and draws 20 - 26 with 6 winners.

Now you may think I'm selling you a story, an exceptional coincidence that happens once in a lifetime. Let's see if this is the only case of clusters. Here is another prediction method: digit growing trend computed from 15 draws, the same data, the same time frame.

 
 

The same story as with repetitives: 19 wins in 30 draws, 2 winning clusters, 9 & 6 wins long respectively. And this is not all. Virtually all prediction methods analysed had such clusters, many of them - at the same time. So, even if I did not pick the best method more that likely I would still end up as a winner.

For this system you don't play straight wheels - way too expensive. But wheels for boxes, either for singles or doubles have only 56 combinations each. This is manageable, particularly, if you have a very reasonable expectation to win. Winning clusters, by being so common and easy to see, make it difficult not to win but to loose. That's why I put in the title - "follow the winner", because, at times, winning is practically a sure thing.

There is another issue you need to be aware of. Statistically 1 in 6 box plays turns out to be actually a straight hit. So you can play box wheel but target a straight. With 19 hits in 30 draws straights will come, regardless of odds. This is particularly valid when it comes to doubles. Doubles are much easier to hit than singles (you need to match only 2 digits, not 3 as for singles) and not unusually double draws tend to cluster once in a while (I've seen 10 doubles in 14 draws).

Winning streaks (or clusters) give you 2 extra benefits: confidence that you can win by choice and not only by luck and thus allow for planning strategies. Some lotteries allow for increasing the bet for each play so you can win more than standard prize. Perhaps it's time to try it. Basically you don't increase your risk significantly but you may double your prize.

Winning clusters can form even with 7 digits wheeled. Repetitive numbers had 2 such clusters: 4 in a row & 4 in 5 draws, while growing trend 4 out of 5 and 5 out of 6. The clusters were much shorter and spread apart thus much harder to predict.

Backtracks are also clear that you can't rely only on one prediction method. Constant monitoring for forming clusters is necessary. You have to do it with every draw - or risk missing good opportunities for earning some cash.

And this is how you can make money (not just win) in any Pick 3 lottery. FOLLOW THE WINNER!