Expert Lotto has a tool which can automatically predict the future winning numbers, future statistical parameters and future values for the History feature. It is based on comparing of several latest draws with the rest of winning numbers database, so called pattern matching.
How does it work?
The prediction algorithm is very simple and applicable to any sequence of numbers. Let's have the following sequence of numbers in the range of 0 to 9:
Position: | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 |
Value: | 1 | 2 | 3 | 0 | 9 | 8 | 5 | 1 | 6 | 8 | 3 | 9 | 9 | 8 | 0 | 7 | 2 | 7 | 9 | 1 | 3 | 4 | 3 | 5 | 6 | 1 | 6 | 2 | 7 | 9 | ? |
Now we are looking for number that will follow in the sequence at position 31. We will start with four latest values at positions 27 to 30 (four latest drawings) - 6,2,7,9 - and will be looking for a similar sequence at preceding positions (previous draws). Such a pattern can be found at positions 7 to 10, where the values are 5,1,6,8. The patterns are not totally equal but they have the same 'character' - increase by 4, decrease by 5, increase by 2, so it is a perfect match. Pattern from positions 7 to 10 is followed by value of 3, so in this case the prediction result is value 4, because the pattern from the end of the sequence has all values smaller by one.
We can also compare bigger patterns. Let's pick the last five values at positions 26 to 30, which are numbers 1,6,2,7,9 and we will be looking for a similar pattern in previous positions. In this case the best match is at positions 15 to 19, where the values are 0,7,2,7,9. Value at position 20 is 1 so in this case the prediction result will be number 1.
The
prediction results from the examples above are different from each other.
So the next problem is finding the optimal pattern size for matching to
achieve as credible prediction results as possible. The automated predictor
in Expert Lotto solves this problem by running multiple predictions for
different pattern sizes from a given interval (see prediction input parameters
below in this chapter) which are then back-tested on several latest values/draws
- the number of back-testing draws is another input parameter for the
prediction process, see below.
The predictor starts with the lowest possible pattern size and evaluates
only values at positions 1 to 29. The actual value at position 30 is then
compared with predicted value. In the next step only values at positions
1 to 28 are use to predict value at position 29 which is then compared
with actual value. Prediction for sequences 1 to 27, 1 to 26 etc follow.
This the predictions are back-tested on e.g. five latest values/draws.
The pattern size is then incremented by one and values
for positions 30, 29, 28, 27 and 26 are predicted again and again compared
with the actual values at those positions. The optimal pattern size is
the lowest size where the most test predictions matched the real values.
This optimal pattern size is then used to carry out the actual prediction
for position 31.
Note: After running the predictions in Expert Lotto you can see the optimal pattern size and the percentage of correct testing predictions in a tooltip of predicted values. The higher the percentage, the more credible the predicted value is.
Note: It is possible that the predicted value can be out of the expected range of input values sequence. For example prediction for Sum root can produce value -2. It happens for example when the sum root for the latest draw is 1 and the predictor expects a big decrease in the next draw. In such cases it is possible either to take into account only the predicted direction - increase or decrease - and filter the Package as such, or ignore such a prediction as not credible and do not base your strategy for the next draw on it. The prediction results can also display N/A which means that the predicted value is outside of interval Average_Value - 2 * StdDev to Average_Value + 2 * StdDev.
Automated prediction in Expert Lotto is available in Winning Numbers Statistics and in Summary table in WN History page. In both cases clicking Predictions button opens predictor window. The predictions start after selecting values to be predicted and entering input parameters (minimum and maximum pattern size, the number of draws for pattern size selection). Prediction results show as another row in the table:
The figure above shows prediction results in Summary table carried out according to the input parameter shown in the picture. We can see for example that the predictor expects decrease in column "0" to value 328 and only a slight increase is predicted for column "-2" to value 1203.
Note: Every lottery has a different history of past winning numbers therefore we recommend experimenting with predictor's input settings to achieve the best prediction results. You can easily verify the efficiency of predictions by hiding the latest draw in your database and then comparing predicted results with the actual results for the hidden draw.