The home of intelligent horse racing discussion
The home of intelligent horse racing discussion

Systems Building

Home Forums Tipping and Research Systems Systems Building

Viewing 9 posts - 1 through 9 (of 9 total)
  • Author
  • #1044
    dave jay
    • Total Posts 3386

    There seems to be a bit of debate going on at the moment of how a system should be built. So thought I would post up my slant on it and maybe get some new ideas or suggestions from other members.

    These are some of my views and techniques that I’ve found useful in the past.

    True Chance and Systems Building (My Logic).<br>Symmetry – is the name given to the number of times an event will occur over a number of frequencies, or its true chance. This is sometimes known as Big Number Mathematics.

    In relation to the liklihood of Tails being the outcome of spinning a coin. The true chance is 50%. Each spin is not notfluenced by the previous spin or the next one. The spins of the coin are unrelated.

    Spins         Outcome          Frequency<br>10              6                      0.600 (60%)<br>100            48                    0.480 (48%)<br>1000          510                  0.510 (51%)<br>10000        5070                0.507 (50%)<br>100000      50400              0.504 (50%)

    The example above shows that out of 100,000 spins of the coin, tails occurred 400 times more than heads, without breaking the laws of averages or going against the true chance odds. So, the bigger numbers of spins give a better estimation of symmetry, or true chance odds.

    So when building a system of your own, you first have to determine the true chance of your selections, or their frequency. Plainly speaking, this means that systems built on small samples of data will be less accurate than ones built on large amounts of data.

    In my experience you need at least 500 outcomes before a picture starts to emerge.

    An example;<br>An aspiring punter tracks a system that throws up 5 bets a day.

    For 100 days he tracks the system to get his 500 results.<br>He finds that his odds-on shots are pulling the system down, so he eliminates them from his selection criteria. They are 40% of his selections.

    500-40% = 300 results left.

    He then needs another 200 results to make up his 500.

    The system is then only throwing up 3 bets a day (40% eliminated)

    So he needs to go for another 200/3 = 66 days to get back up to his 500.

    So, after 166 days he has 500 results and sees that the long priced shots are also a burden. So he decides to eliminate everything forecast at 9/1 or more. They are only 20% of his selections.

    500-20% = 420 results left.

    He then needs another 80 results to make up his 500.

    The system is now only throwing up 2 bets a day (20% eliminated from the first total)

    So he needs to go for another 80/2 = 40 days to get to his 500.

    After 202 days our dedicated punter is ‘starting’ to get a general picture.

    Any other modification will require the same amount of rework and data gathering. Building a system can take up to a year IMO and at the end of it you may decide that it doesn’t work!

    I think this is a hard lesson to learn, but there is wisdom in it and it can be used to the punters advantage. The way I see it, there is no rush, racing isn’t going anywhere. Better to trade a system in this way than to trial it with your own real money and come to the same conclussion. If your personal circumstances are pushing you towards gambling, then your psychology is all wrong, there are no short cuts.

    Using tools like RacingSystemsBuilder will help but you still need to do this IMHO, because you need to build confidence in what you are planning to do.


    • Total Posts 554

    Interesting DJ I have a love of mathematics saddo that I am – and I can certainly confirm that a large No of repetitions are necessary to see a truer proximity to symmetry. But beware – not for nothing is ‘The Law of Averages’ also known as ‘The Gamblers Fallacy’.

    You will only arrive at a more representative %age figure and not true symmetry as you have indicated.

    This is where the problem lies – because the bookmaker is playing in every race he reaches a representative sample much more quickly than you the punter.

    It’s probably easier to understand using roulette as an example. Let me say straight away that roulette or any casino game with fixed odds is a mugs game for everyone but the house, however because there are a fixed No of numbers, symmetry (how often something should happen in theory) is easier to work out.

    In the UK there are 37 Nos on a wheel 0 plus 1 – 36. Zero is green 18 Nos are red and 18 black. I wont complicate it any more for this example. The Law of Averages (which by the way isn’t a law of physics at all) states that out of each 37 spins each No should appear once – one zero – 18 reds and 18 blacks.Now if someone is backing reds continuously they will win 18 times, lose a full stake 18 times(when black comes up) and lose half a stake(when zero appears) thus IN THEORY they will lose 1 unit every 74 spins. BUT because the punter is dipping into the long continuous line of spins and taking (comparitively speaking) a very small sample it is likely that he will get a very unrepresentative set of results.

    This is where so many punters fall down they read in the Racing Post that a certain type of race has a 34% strike rate for favourites and they rub their hands in glee. Better than 1 in 3! They rush off and start betting on this type of race on the favourites in ever increasing stakes only to go broke in 10 days. Back to the casino red should appear 18 times out of 37 almost 50% of the time but any pit boss will tell you that runs of one colour or the other of 16 or 17 are commonplace. Likewise the aforementioned Favs – the punter forgets that there may be 18 races of this type and 14 winning favs – but somewhere along the line there may be losing runs of 20 or more to compensate for it.

    This is probably the last time I will post before xmas so can I take this opportunity to wish you DJ and indeed all forum members a happy christmas and a prosperous 2003.

    dave jay
    • Total Posts 3386

    That’s a good post snowman!

    You make a very valid point about losing streaks here is a staking plan I would like you (and others) to take a look at. I’ve had this one for a while and it was posted earlier this year in ‘Gummy Racing’. It’s not there any more. Personally speaking, this is one of my favourite staking plans and can be very useful when assessing results data.

    ‘The Retirement Staking Plan’

    This plan is based on the SP of the winners thrown up by any system that produces level stakes profit.

    To set up this staking plan you have to do the following:<br>

    • <br>
    • Calculate the Average Price of your previous winners.<br>
    • Calculate a Divisor.<br>
    • Invest a regular percentage of your bank.<br>
    • Set a reasonable target.<br>

    <br>Firstly, calculate the average price of your previous winners. You will need at least 50 winners to get this figure.

    Add together the return SP figures and divide the total by the number of winners. Such as:<br>7/4 + 4/1 + 5/1……<br>1.75 + 4 + 5……………<br>I always deduct the two highest and two lowest SP returns to ensure the Average Price is not skewed by fluke results.

    This is the key to this plan because this determines, the divisor and when to bring in your safety device.

    Let’s say that the average price is 3/1.

    The best way to show you this plan is in the table below. Firstly, we have a divisor, which is double the average starting price winner of the system you are using. So, if the average price winner is 3/1 then the divisor will be 6.

    We have a divisor of six and the first bet will be 1% of the bank.

    Now to establish a target, multiply the divisor by your first bet, 6 x 100 = 600. So your divisor is six, target 600, first bet 100, which is 1% of your bank, all losses are added to your target, and if you go six without a winner, then you start to increase your divisor by one after each losing bet.

    This is what happens after a run of losses:

    DIV   Targ     Bet   P/Loss   Result    Bank<br>6       600      100   -100       L           9900<br>6       700      115   -115       L           9785<br>6       815      135   -135       L           9650<br>6       950      158   -158       L           9492<br>6       1108    185   -185       L           9307<br>6       1293    215   -215       L           9092<br>7       1508    215   -215       L           8877<br>8       1723    215   -215       L           8622<br>9        1938   215   -215       L           8447<br>10     2153    215   -215       L           8232<br>11     2368    215   -215       L           8017<br>12     2583    215    +860    W4/1     8877<br>8       1723    215    -215      L           8662<br>9       1938    215    -215      L           8447<br>10     2153    215    1290     W6/1     9737<br>6       863      145    435       W3/1    10172

    As you can see, losses were added to the target, and when we went to six without a winner, we increased the divisor, if you do not the bets get out of hand and jeopardise the bank.

    If you look at the bank after six losers 9092, it could still withstand another 42 losses without a winner. That’s not because we started with a bank of 10,000, it would be the same if your bank was 1,000, as your first bet would be £10 instead of £100. This plan can go for 48 bets without a winner.

    So, back to the table. After the 4/1 winner, the target is reduced from 2583 to 1723. You look back then to where you were in the staking plan  were your target amount was close to this figure. You see that your divisor was 8 and start staking again from this part in the plan.

    I normally, skim the bank when it goes into profit and track the staking plan’s progress from a table, rather than getting all technical at every bet.

    (If anyone doesn’t know how to set this up in their ‘spreadsheet’ on the pc. send me a PM and I’ll give you the details).

    What I like about this plan.

    IMO this is better than level stakes because it acknowlegdes the fact that any system is subject to losing runs.

    It is good for calculating your longest losing run. As in the example above the average price was 3/1, this means that the system must have a strike rate of 25%. I reckon that you should look at losing runs as being twice the size of two losing frequencies in ten, before your in trouble.

    In this case 2.5 win in ten and 7.5 lose.

    So my losing frequency is 7.5 (x2) = 15.

    Well, that’s what I think anyway! And Look forward to your critique!

    All the best for the New Year snowman and all of the pals in the systems workshop!

    • Total Posts 13

    Hi there,

    Hmmm maths not my best subject!

    Can you explain what the purpose of the ‘Target’ figure is? The first bet needs 6/1 for that, is that what the actual bet would have been? Do you aim to achieve that (600 say) before you cream off the profit as it were and begin again?

    Also, how does the bet go from 100 to 115?

    I am sure I am just being totally thick and missing something blindingly obvious.


    dave jay
    • Total Posts 3386

    Hi stephen12000 and welcome to the forum.

    The example below shows a losing run of 11 and a break even situation, but the plan still returns a profit of £172 after turning over £2988. In real terms this is a profit of 6% on investment.

    The Target and Divisor should be looked at together and not independantly from one another. The Divisor is the sum total of all of the system’s winning SP’s divided by the number of them. In this case the average price is 3/1 (meaning that the system will have a 25% strike rate if it breaks even from the; calculation 1 / 3+1 = 0.25). By doubling this to 6 there should be 1.5 winners in every 6 bets from the calculation; 6 / 100 x 25 = 1.5.

    Therefore, the plans aim is to stake 1% of the bank at the first bet, so the Target figure is 1% of the bank multiplied by the divisor, for the first bet, which in this case is £100. The second bet is the Target plus the lost stakes from the first bet divided by the divisor, which in this case is £115. The stakes are then increased in this way until there are six consequetive losers when the safety devise is kicked in and the divisor then goes to seven.

    So, the Target figure, is not my target profit, as such, but the figure used to control my stake value in relation to previous losses and starting stake of 1% of the betting bank.

    The profits from the staking plan are counted every time the bank goes above it’s original starting value. When you take profit is entirely up to the individual, you may decide to lock profits into the plan and build your starting bank, taking profits yearly, or skim the bank monthly and put any profit in your pocket.

    I hope that explains things a little more clearly.

    • Total Posts 5824


    Could you send me details of how to set up up a spreadsheet for your staking plan so I can put through some figures?

    I am afraid I am a complete "dunce" when it comes spreadsheets.

    Regards – Matron<br>:cool:

    • Total Posts 13


    Thanks for taking the trouble to clarifying things.

    It seems a little complex at first, but now I see better. It is the kind of plan I have been trying to figure out and use for myself but getting nowhere.

    Thanks for the welcome too, it seems a friendly board


    dave jay
    • Total Posts 3386

    No, problem Stephen,  

    This type of staking plan doesn’t rely on a high strike rate but on level stakes profit. So it suits a strike rate of around 20%. In the example there are only 3 winners from 16 bets (18%) and although there isn’t a profit there also isn’t a loss.

    Matron here is an introduction to spread sheets and how to set up the plan shown. This works in basic spread sheets and excel. It takes a bit of practice but once you get used to it it’s quite easy. You might want to print this off before you start.

    Open up the Spread sheet function.<br>The sheet is split into a series of boxes called ‘Cells’.  Across the top of these are letters naming each ‘Column’ and down the side are numbers naming each row. So, cell ‘C3’ is the cell on row 3 in column C.

    Values in each cell can be calculated against one another using symbols;<br>/ is divide by<br>* is multiplied by<br>- is minus<br>+ is add<br>^ is to the power of

    Each formula starts with the following; =sum

    Example, try this,<br>In cell c3 type a 3 in cell d3 type a 4. in cell e3 type =sum(c3+d3) and then press enter. The value shown in e3 should be 7.<br>Click back on the cell with the formula in it and you will see that the formula appears in the formula bar at the top of the spread sheet. To amend the formula click in the formula bar. Try to amend the formula using the different symbols shown above, in lieu of the + sign. Do this until you are comfortable with what is going on.

    To set up the staking plan as shown;

    Type in the following information into the following cells;

    Row 3 only<br>Column<br>A – Sequence<br>B – Divisor<br>C – Target<br>D – Bet<br>E – P/Loss<br>F – Result<br>G – Bank<br>H – 10000

    These are your column titles, now populate the columns,<br>In column A fill in 1 to 17, this is the series of bets.<br>In column B, manually enter the values as shown in the earlier post.<br>In column F, enter results L for Lost and winning odds in numbers 6 for 6/1, etc.<br>In column C, for your first target (c4) type =sum($h$3/100*1)*b4. In the next row (c5) add the formula =sum(c4+d4) copy and paste this to every row after each losing bet. In column c row 16 add the formula =sum(c15-e15), copy and paste this to each row beneath each winning bet.<br>In row D, add the formula =sum(c4/b4). Copy and paste this into all cells in this column.<br>In row E, (e4) add the formula =sum(d4-d4)-d4, copy and paste this to each row that has a losing result. In (e15) add the formula =sum(d15*f15), copy and paste this in every row that has a winner.<br>In column G, (g4) add =sum(h3+e4). In (g5) add the formula =sum(g4+e4), copy and paste this into all subsequent rows.

    By now if you’ve done it right and I’ve explained it properly, you’re formula cells will be full of numbers running out into decimal places. To sort this, highlight each column you are using along the letter at the top, so each column is black. Click on the format buttong at the top of the screen. Click on number and adjust the decimal places to ‘0’. Leaving the columns highlighted click on format, column, autofit width. And that’s it.

    Note that ‘$’ (dollar sign) is added to the formula cell (c4), this pins the formula to always get it’s data from that cell when you copy and paste it. Copy and paste (c4) to (c20). This starts the betting again when the target drops below 600, providing you have entered the divisor.

    Have a play around with this and let me know how you get on Matron. As I said before it’s not as complicated as it looks and once you’ve decide how your cash flow will work it’s just a matter of copy and paste.<br>

    (Edited by dave jay at 8:51 pm on Dec. 30, 2002)

    • Total Posts 5824

    Thanks for that Dave.

    I will print off and have a play.

    Thanks again.

    Regards – Matron<br>:cool:

Viewing 9 posts - 1 through 9 (of 9 total)
  • You must be logged in to reply to this topic.