# Lotto Strategy Around Filter Averages

Written by Nik Barker on August 09, 2002.

A few weeks ago I tried a strategy thus:

1) Look at the W6.1 report and count down to line 15
2) If 10 or more of the last 15 draws' filter values for a particular lottery filter are ABOVE the median, then 'expect' the next draw that the value of that filter will go down. Hence, set that filter's value to MAX = (median+1)
3) If 10 or more of the last 15 draws' filter values for a particular filter are BELOW the median then 'expect' the next draw that the value of that filter will go up. Hence, set that filter's value to MIN = (median)
4) Do the same for the other three W6 reports

I don't know where that came from, but it seemed worth a try. I created also a LEAST6 file covering 12 least pairings for each lotto number and added this to the top of the D6 file remembering to put the figure '588' in place of the second zero in the top line of the ST6 file.

I ran WHEEL632 (choosing the '6 out of 6 guarantee') using the above strategy.

It generated a total of 58 combinations. These were generated with an up-to-date history D6 lotto results file and I had time to produce the play-slips and get down to the shop and put those lines on for that night's draw. I decided against it. Fifty-eight quid is a lot of money to me right now…and it was too much to risk, just on a hunch. However, the hunch would have paid off. Those 58 combinations scored:

7 x 3/6
2 x 4/6
1 x 5/6

in that very night's draw. (About £1,700 worth of winnings !) I could be kicking myself right now, but I'm not. I know what I'm like. If I'd played with real money and I'd have won real money, then the temptation would have been so great to really believe that I'd 'found something', that the theory was 'valid' without question, and probably before long I'd have lost all those winnings by playing the theory to death before the next win. I also think I've found something even better!

Now don't get me wrong – I do believe there's something in it. Let's just look at it for a moment.

Each of the filters has a median. Taking the UK Lottery, each filter's median is the middle value of a string of (currently) 691 values. If the next 6 draws produce 3 values above the median and 3 values below the median, then what will happen to the median? It will stay exactly where it is, the same value it was before those next 6 draws.

So then what happens if the next 50 draws produce values that are all above the median for a particular filter? Well it's quite possible that the median value will change compared to what it was 50 draws previously (it depends on how far the median value carried into the upper-half of the string of values).

This was the basis of the 'theory'. If out of the last 15 values for a given filter, 10 or more have been above the median, then I was 'expecting' the value to go lower, to restore the balance if you like, of that median value. However, if values keep coming out above the median and only above it, then the median value will eventually increase. I don't know why I picked 15 draws to go back, I'm not sure. Perhaps there is an optimum number to go back and an optimum number of Above-vs-Below-the-median to look at. But I haven't discovered that yet.

I decided the only way to see what treasures might be lurking in the w6 reports was to write a bit of code that would analyse the W6 reports, so I wrote that bit of code. I can now set it to look back a certain number of draws – whatever I decide. It will then look at each filter's values for that range of draws.

I decided to sort of turn the 'theory' on its head. I thought, 'I'll use the Average values of each filter, rather than the median, just to see what happens. AND MOST SIGNIFICANTLY I will inverse the expectation. IE, if 9/10 or 13/16 (or whatever) of the previous filter's values have been ABOVE the Average, and then the next value ALSO was higher, then I will print a 'YES' on the output file for that filter. In other words – to contradict the title of the great Alan Parsons track 'What goes up must come down' – what goes up, keeps going up…..and vice versa.

When I ran the program I could not believe my eyes! It produces a table like the one below…but let me just explain the following. The program reads EACH W6 file for each layer. It then takes each line in turn and counts back 7 lotto draws (this seems to be the optimum for analysing against the AVG) for each filter. It compares these values with that of the Average value for each filter. Where 6/7 or 7/7 of the previous 7 values were ABOVE the average AND the next value ALSO went above the average, it puts a 'Yes' in the relevant layer and for the relevant filter. Likewise if 6/7 or 7/7 of previous 7 values were BELOW the average AND the next value was ALSO below the average then it also puts a 'Yes'. If the opposite is true, it puts a 'No'. If however, there was more of an even split of values over the previous 7 draws like 5/2 4/3 (above/below the average), then it leaves this filter blank. I should also say that for reasons that it's not possible to set a MAX value for FIV 'Yes's and 'No's in these columns DO NOT COUNT towards the summary figures for each draw (nor at the end of the output file)…ie if you look at Draw No 1 below and you add up all the 'Yes's and 'No's you would find you have 8/11 = 'Yes', but I have not counted the FIVr 'Yes' in Layer3, so it becomes 7/10 = 'Yes'.

```Draw No:  1

ALL  POT  TWO  THREE  SUM  FOUR  FIVs  FIVr  BUN4  BUN5  BUN6

Layer1                          Yes   No
Layer2                          Yes   No
Layer3     No             Yes   Yes                    Yes   Yes
Layer4                          Yes              Yes

Draw No:  2

ALL  POT  TWO  THREE  SUM  FOUR  FIVs  FIVr  BUN4  BUN5  BUN6

Layer1     No                   Yes  Yes
Layer2     No                   Yes  Yes
Layer3    Yes             Yes   Yes                    Yes   Yes
Layer4                          Yes              Yes

10 / 12 = 'Yes'
Draw No:  3

ALL  POT  TWO  THREE  SUM  FOUR  FIVs  FIVr  BUN4  BUN5  BUN6

Layer1                     No   Yes  Yes
Layer2                     No   Yes  Yes
Layer3                    Yes   Yes                    Yes   Yes
Layer4                          Yes              Yes

9 / 11 = 'Yes'
Draw No:  4

ALL  POT  TWO  THREE  SUM  FOUR  FIVs  FIVr  BUN4  BUN5  BUN6

Layer1                          Yes  Yes
Layer2                          Yes  Yes
Layer3                    Yes   Yes                    Yes   Yes
Layer4

8 / 8 = 'Yes'
Draw No:  5

ALL  POT  TWO  THREE  SUM  FOUR  FIVs  FIVr  BUN4  BUN5  BUN6

Layer1                          Yes
Layer2                          Yes
Layer3                    Yes   Yes                          Yes
Layer4

5 / 5 = 'Yes'
Draw No:  6

ALL  POT  TWO  THREE  SUM  FOUR  FIVs  FIVr  BUN4  BUN5  BUN6

Layer1                          Yes
Layer2                          Yes
Layer3                          Yes                          Yes
Layer4

4 / 4 = 'Yes'
Draw No:  7

ALL  POT  TWO  THREE  SUM  FOUR  FIVs  FIVr  BUN4  BUN5  BUN6

Layer1                          Yes                           No
Layer2                          Yes                           No
Layer3                          Yes         No
Layer4                                      No

3 / 5 = 'Yes'
Draw No:  8

ALL  POT  TWO  THREE  SUM  FOUR  FIVs  FIVr  BUN4  BUN5  BUN6

Layer1               No         Yes                     No   Yes
Layer2               No         Yes                     No   Yes
Layer3          No              Yes
Layer4               No               No   Yes          No

5 / 13 = 'Yes'
Draw No:  9

ALL  POT  TWO  THREE  SUM  FOUR  FIVs  FIVr  BUN4  BUN5  BUN6

Layer1              Yes         Yes                    Yes   Yes
Layer2              Yes         Yes                    Yes   Yes
Layer3         Yes              Yes
Layer4     No       Yes                                Yes

12 / 13 = 'Yes'

..
Draw No:  684

ALL  POT  TWO  THREE  SUM  FOUR  FIVs  FIVr  BUN4  BUN5  BUN6

Layer1
Layer2
Layer3               No                                            Yes
Layer4    Yes             Yes                                Yes    No

4 / 6 = 'Yes'

Total number of draws where ALL filters were 'Yes' =  80

Total number of draws where all-but-one filters were 'Yes' =  117

Total number of draws where all-but-two filters were 'Yes' =  131

```
You can then set the 'Yes' filters to MIN = (average), or MAX = (average+1) depending on whether it is going on an 'upward' or 'downward' trend.

I realise that averages change as draw histories get longer. And it is true to say that in the last draw analysed 684 (the first draw ever) the program IS comparing these filter values with the CURRENT averages. Also using these filter settings of themselves still produces around 10,000 combinations (though without BEST6 or LEAST6), but in a test, with one of the draws for where all the filter values were 'Yes', I ran WHEEL632 to 1000 combinations. When compared with what the next draw would have been, those 1000 combinations produced £4,500 - £5000 worth of winnings.

That is trying it raw though. IE without BEST6 or LEAST6. The 'theory certainly isn't proven. If there's nothing in it then fine…it sure was fun finding it though and I haven't dismissed it yet! There is certainly a lot more work I need to do to refine it and see if it can work over prolonged periods, but I should add that in the first example I gave you above (the one where 58 combinations were generated) LEAST6 CONTAINED(!!!) some of the pairings that came out next draw…AND some of the filter values would have showed up as a 'No'…and yet it still produced £1,700 worth of winnings.

Could be worth looking into more.

Cheers
Nik Barker

Ion Saliu's Note
• The strategy is in accordance with the Fundamental Formula of Gambling. It can be used as a casino gambling strategy, where there are fewer filters or parameters.
WHEEL632 as well as all my software is re-writing itself at this time. Nik and others informed me that the SIM-6 files have sometimes the 6 winners. Unfortunately, that is a side effect of a much-improved randomised function. The lotto random generator is so good now that it beats random expectation way too easily. In other words, it matches real life drawings a lot more often than I desired.
I use the following method to create efficient SIM files. I use a normal D6 file with all real draws on top; say, 500 real draws. I generate a new SIM file, eliminating FivR and FivS from the 500 real draws. Still, the new SIM combinations will hit the jackpot (undesirably), but less frequently. Also, I might consider loosening a few inner filters.
IS