I used to play Pick 3 lottery for some time but generally with little success (doesn't it sound familiar to most folks?). At first I observed the numbers for a while and then picked a set of my personal favorites. I played them over and over again. But I seldom had any success with them - they did not want to come up, at least not in sequences I used for playing.
I also used to play Lotto 649 before and experience from it gave me some insights. Personally I prefer prediction methods based on statistics on current numbers drawn. I used them in Lotto 649 and applied them to Pick 3 as well. They are rather simple, easy to calculate (even manually, but preferably by computer which is much faster and less error-prone) and can be easily backtracked.
I selected the following methods for predicting numbers: hot, cold, repetitive, due to hit, overdue, and growing in popularity.
While I could see differences in efficiency between these prediction methods they all had one thing in common - they all fluctuated, sometimes they won and many times they didn't. Similar to what many people expressed on LP.
So I conducted backtracks on all my methods. The results were a big surprise to me. They showed me that the reasons why I was not winning lied not in the methods themselves but in the way I was using the predicted numbers. In other words, it matters much more HOW I use my numbers rather than WHAT my prediction method is.
Backtracks showed me clearly what I SHOULD NOT be doing when playing Pick 3.
1. Using one method only.
Backtracks confirmed without a doubt that all predictions methods fluctuate significantly in efficiency. Using one method exclusively put me outright in the "looser" category. I could not win consistently with one method only, at least not for any extended period of time. I had to monitor all of them for performance (which was expressed by the number of wins for the same, comparable, period of time).
One of the surprises was that, at one time, cold numbers outscored hot numbers by a significant margin, which sounded completely illogical, but it happened. Hot numbers were always (naturally) my personal favorite, so I learned a valuable lesson - no favorites. Play only what's winning now.
2. Trying to time the numbers precisely..
Many times I was playing a small selected group of "favorite" numbers (usually 3 or 4) arranging them in varieties of combinations that I was sure would win. They very rarely did. It is basically impossible to regularly predict with any degree of accuracy sets of 3 numbers - for example, the numbers 750 will be drawn today.
Backtracking clearly confirmed that: all methods tested yielded the same perfect score: zero.
3. Mixing goals.
In Pick 3 there are 5 win levels: straight singles, straight doubles, box singles and box doubles, and triples (I ignore the last one). To increase winning probability I was often playing for all of them, a mixed bag of combinations of my favorite numbers.
But it was like trying to hit 4 different birds, each flying in different direction with different speed and pattern - with one stone. It basically meant throwing the stone high in the air and hoping that one of the birds would accidentally collide with it. It happened on few occasions but rather seldom, with no consistency.
The backtracks clearly indicated that, in order to be more successful I have to target precisely what I want to win and aim at one target at a time. Either straight or box, and then either singles or doubles. No mixing combinations.
And the main reason for targeting was that I wanted to play more than 3 or 4 numbers at a time.
4. Playing insufficient numbers.
As it turned out playing insufficient numbers was one of the main reasons for my failures.
But if you want to play more than 3 numbers you have to wheel them. (For unfamiliar, wheeling means distributing numbers into combinations that cover all possibilities so no matter which of your selected numbers are drawn and in whatever sequence you win something).
Since I could not find any reliable information on Pick 3 wheeling anywhere I designed my own wheels. Actually, it was not that difficult. What turned out was that wheels are strictly type-of-play dependent. Which means they are different for straight and box plays, and within these - for singles and doubles. And that's a big reason for targeting, because these wheels are not exchangeable.
More numbers for playing requires bigger wheel to cover all combinations. So I experimented with different number of numbers for different plays.
Backtracks were pretty consistent. From 3 numbers to 6 (inclusive) the results were quite similar - usually zero, seldom more than that. Playing this way I would not spend much money but I would not win much (if anything) either.
When I started to wheel 7 numbers the methods began showing some signs of life. I would win something at least once in a while. Still, not as frequently as I wanted.
The fun began with 8-numbers wheeling. The wins became quite frequent and consistent. Although there were fluctuations between methods the consistency was there for all of them. Also, while wheeling 8 numbers does require quite a number of bets, if you win frequently you can still turn a profit, at least from time to time. Considering the cost-to-win ratio 8 numbers seem to be optimal.
The conclusion is simple. If I want to win in Pick 3 more consistently I have to wheel more than 3 numbers. There is no alternative.
5. Constantly targeting singles.
No question that box singles appear an easy target. Most of my plays were aimed at them. But, unfortunately, as the backtracks demonstrated, money is in doubles. Here is why I think so.
In order to correctly predict a double we need to predict only 2 numbers. Odds for that are 10 x 10 = 1 in 100. But to predict a winning single we need 3 numbers: 10 x 10 x 10 = 1 in 1000. No question which odds are better.
Moreover, doubles pay more money for a win, usually twice as much as singles. Financial benefit obvious.
Here is what my backtracks told me: when wheeling 8 numbers, singles would win only approximately between 30 -50% of draws. For doubles this ratio would be 70-90%. These ratios apply to all prediction methods - because odds are simply better for doubles.
6. Failing to monitor the draws.
It never occurred to me before to watch the actual draws for types (levels) of win. Surely, I was watching the numbers for patterns but why types of wins (singles or doubles)? It did not seem relevant. But it actually is - very relevant. Without watching this I cannot target effectively. It's like shooting with eyes closed.
72% of all possible winnings are singles, doubles constitute only 27%, roughly 3:1 ratio. Does not look good, does it? But these are only statistical averages. Actual percentages, particularly in a short period of time (2 weeks, for example) may be totally different. I have seen split 50-50 between singles and doubles in a period of 12 days, or 60-40 (for singles) in the whole month. That does not look that bad. In fact, these are the situations to take advantage of.
Win percentages have to be monitored all the time. This is very important for targeting doubles. Can I predict 100% when doubles start coming? Of course not. But if doubles, in a short time, (again, 2 weeks for example) are grossly underplayed (I've seen their percentage drop to 13%) it is fairly safe to assume that they will start coming back. And that's the time to play them - because of their much higher efficiency ratio and higher payout.
Moreover, not unusually, win types come in streaks: several singles or doubles one after the other. If there were no streaks for a long time chances are one may appear any time.
To sum up my observations:
The backtracks told me clearly that it is far more important HOW I play my numbers than WHAT these numbers are. Targeting, wheeling and draws monitoring, all combined, take precedence far ahead of any single prediction method, no matter how good such a method may appear. The "perfect" prediction method simply does not exist and it is a waste of time to look for one.
So the final lesson is: backtrack before playing. It's a must, not an option.