How to find an Edge!
 

Part 1.
This thread will be the first in a series of tutorials with the aim to find an edge in racing statistics. You could use it for Dogs, Trots or Gallopers. It is aimed at newbie punters, without the benefit of a searchable database of past results, and will arm you with both the discipline and techniques to improve your win strike rate, be it Backing or Laying. I am passing on everything I have learnt in the past six months since deciding I wanted to make an income out of punting.

You need to decide on a Price Range that you want to investigate. It might be odds/on, mid range or longshot prices. In my examples I am looking at 1st favourite odds/on gallopers, price range $1.00 ~ $2.00. The techniques I will pass on will empower you to exploit any price range of results and find if there are any profitable niches.
These niches are not found looking at a racebook or the racing sports page. Once armed with the knowledge gained from your examination of results you will be able to write your own filters and rules to guide you through each day of your punting life.

Firstly you need to gather at least 300 race results that included a horse within your price range. You can source all this information from Betfair & Racing & Sports, I have attached an excel spreadsheet view to guide you. The data you'll need:
Date of race
Venue
Race No.
Number of runners
Horse No.
Horse name
Betfair Start price
Last 3 starts recorded as: if first start 'FS', if 2nd start 'SS', if returning from spell 'RFS', if 2nd or more times returning from spell '2RFS'. if horse has had a min of 3 starts in current campaign then record numerical finish position of last 3 starts.
+ or - weight change of rider since last start in kgs
+ or - distance change since last start in metres
Class change since last race recorded as either 'up in class', 'down in class' or 'no' change in class.
Result of race recorded as 'Won' or 'Lost'
Maiden race or not?
2nd Favourite's Betfair Starting price
Win strike rate as a %
First up strike rate as a %

In the 2nd part of this tutorial we will begin to analyse what we have recorded for profitable filters.

Part 2
OK you've gathered your min 300 results. Now from the image attached you'll see that I have a sample of 11 results. I can tell you that from my stats the numer of races WON by an odds/on favourite is about 54%. So to simplify things for this tutorial, it works out to be about 6 winners to 5 losers hence the sample of 11 races I've shown.

Under the results I've sorted and displayed races "Won" at the top and races "Lost" underneath. You don't have to do this but it helps visualise what's going on. Now on the right of the sorted group we are going to enter a formula as if we were Backing the horses, and adjacent, a formula as if we were Laying the horses.
At cell Q18 enter this formula and then copy and paste it down to cell Q28:

=IF(L18="Won",(G18-1)*93.5,-100)

At cell R18 enter this formula and then copy and paste it down to cell R18:

=IF(L18="Lost",(100/(G18-1))*0.935,-100)

These formula will calculate the nett profit or loss based on whether our horse won or lost. Now enter =SUM(range) at bottom of the two columns to see the result.

If we were Backing Odds/On favourites we would have LOST $98.89 and
If were Laying Odds/On favourites we would have LOST $65.68.

Most punters don't even bother with this examination, they make a quid one day and it makes them feel good. Then another day they'll lose, have the irrates, but be back for more, absolutely mindless to the workings behind this Price range and none the wiser for their experience.

Still, for those of you that have just done this excercise, it doesn't look like a very profitable area of punting does it?
Well I'm here to tell you that by the time I've finished this tutorial and you go off and put it into practice, you'll be sending me a bottle of Moet every Christmas as a thankyou! .....RP

Part 3

One of the pluses with laying Odds/On selections is that because the odds are less than evens, we get a greater return than our original stake/liability. For the backers they get less back! The downside is that odds/on horses win more than they lose to compensate. This is not always the case, and will be detailed in another post.

So providing the horses keep losing and you keep Laying you're a long way in front of the Backers. Now this can be a double edged sword because if at the same time as we Lay the favourite, we Dutch Back the field, we can double our profit!

Look at the attached image of the Betfair results for a race won by "Pinpoint" @ $1.53 SP. Let's say "Pinpoint" lost, it makes no difference to the prices.

"Pinpoint" loses and our profit for $100 liability is (100/1.53-1)*0.935 = $176.41
At the same time we 'Back the Field' using a Dutching calculator (best done by a bot) that I've shown as well.

The result is a profit of $213.43 less commission = $199.55 on top of the $176.41........A total of $375.96

One qualification, your Dutching bot is placing SP bets so there is going to be slippage of stakes v final prices as the SP prices aren't known till after the race starts. Sometimes it can work in your favour, sometimes against. For the purposes of all future tutorials we will assume that your Dutch bet was successful and work on a total profit of 2 x Lay win as profit.

Hi RP. Very interesting stuff so far (thanks for that) but in your Part 3 you have not told the story of when the odds on fave wins. You then have to pay out on your lay bet AND you lose the $100 stake for your dutch backing bets. Granted you would ONLY lose $200 rather than winning $375 (in this case) so maybe that's where your system beats the odds?

Now you're getting the hang of it. There's still a way to go yet, it'll be worthwhile.
Thanks for recording your interest TO, I'd hate to think no one is interested! if I can help ONE person, it's been worth it.

More tomorrow.....RP

Thanks RP. I make it you are about $11 better off than just laying the odds on fave although does that take into account commission and turnover charges?

You would be up $76 if the odds on horse loses for just laying or up $87.50 for both laying the odds on horse and backing the rest of the field in this scenario. ($187.50 is half of $375 just to even it all out for $100). I hope that makes sense.

Don't be in too much of a hurry to think you've worked out the Edge! The Edge is in turning the tables on the win strike rate!
RP

RP
I am sure that everybody is interested but give us time to digest it and form an opinion. I know that I am.

RP
I'm reading, mind ticking over and will waiting patiently.
Thanks for sharing.
Just thinking I don't use bots but could dutch using fixed odds on other bookmakers site and just lay on the BF exchange.

Thank's Willy, but by the time you manually type in the odds for the field, the race has probably started. It has to be done within a few seconds. You could make enough money to pay for the bot in one or two bets. You need a bot that can place the Back & Lay bets at the same time, all staked correctly in the last couple of seconds before the close.
RP

Gruss + excel I have a trading sheet that does just this, could never find a trading edge to get it to work auto.

I've been away for a good holiday the last few weeks and just catching up reading now. Very interesting posts RP. I'm looking forward to your next instalment in this tutorial series.

Good to hear you're back UB. Missed your "Wise Owl" wisdom. I overcame that bot issue, I emailed you about, with BF Bot Manager Multi Strategies and was able to reverse the order of favourites to Lay.
Cheers........... RP

Part 4

The next step is to dismantle your 300+ selections into "bite size" pieces. Sort all results into the following dividend categories : <$1.30, $1.40, $1.50....$2.00. Once completed sort each category with Winners at top and Losers underneath. The attached image is for the category <$1.30

In the cell at bottom of Final Odds enter "=AVERAGE(range)" to get the average of final odds which is $1.24. Now divide 100 by $1.24 = 80.65%. I call this result the FORECAST WSR or forecast win strike rate. It's what the odds forecast as the chances of the selections in this category winning.

Next count the number of races Won (6) and divide by number or races (7) * 100:
(6/7)*100= 85.71%

So we can now compare the two:
Forecast WSR = 80.65%
Actual WSR = 86%

For Backers this is called an Overlay. In other words, selections in this category have more chance of winning than the dividends predict.

(A word of caution: These differences are the result of a very small sample and should not be considered as gospel) It's the technique that's important, and a very powerful one, once you have a larger sample.

OK back to looking at our <$1.30 category. Despite the 80+% WSR the profit isn't so good at $27.16. I'll try and improve on it, so as well as Backing the Favourite we'll Dutch Lay the Field to try and squeeze out a little more profit. I've added that new column to the right of the Back column. Similarly a very small bottom line profit of $9.29.

All up a total profit $36.45 on turnover of $700 = 5.2% POT

But wait there's MORE!!

You need to be like a forensic detective, look harder at the reasons why there were any losses? We see that the loser was "Up in Class". There was a winner too, but if we introduce a new filter and avoid Backing selections in races where favourite is <$130 and "Up in Class" we get a different result. We get 5 wins from 5 races for a profit of $186.85 and 37% POT.

Now remember back in Part 1 of the tutorial we couldn't make any profit out of this Odds/On group but we've just found a niche that's paying 37% on turnover! More importantly we've just learnt a lot more than "mug punter" will ever know about the long term results of betting on selections under $1.30. That's the payday for you, when that light in your head goes ON!!

So if you feel that based on the sampling, the POT is acceptable, we can write our first filter to apply to all results from now on:

=IF(AND(K#="No",G#<1.30),"B"," ")

So what this says is that if the selection isn't up or down in Class and the Odds are less than $1.30 then "B" Back the selection but if not, blank space. You will enter this filter in a cell to the right of your data. It will be one of twenty filters you will employ to build your "Black Box".

Disclaimer: The criteria and formulas used are not a guarantee of future performance. Please be responsible for gathering your own data to apply these techniques to.

Here's a test for you, to see if you've been paying attention, sitting up the back of the class?

Can you see anything else in the results spreadsheet that sets the two selections that were "up in class" apart from the rest?

???????????????????????????????

Look at + or - Dist, WSR% & 2nd Fav odds. You have to keep looking for that Edge all the time. There's not enough there to be conclusive in this small sample but there's an inkling. That's all you need to sniff out a profit.

Have a Great Day...............RP

Part 5

We now move onto the $1.30 category.Looking at the sorted results we see that there's been more losses than wins.
Calculate the Forecast WSR:

100/1.36 = 73.5%

Calculate the Actual WSR:

3 wins/7 races = 42.8%

So the Forecast WSR is 74% and Actual WSR is 43%, known in the game as an Underlay, a Layers best friend.

Based on our tests we are confident to Lay this category so we do our calculations, to the right hand side, to Lay the $1.30 selections and Dutch Back the Field. Result is a profit of $1,561 on a turnover of $1,400 = 111% POT. A Great Result!!

We check any other conditions that might have assisted a sub group of selections winning or losing, but nothing stands out.
Again beware, this is a very small sample used to demonstrate a technique and should not be relied upon to predict future profit/losses.

If confident that you've found an Edge, write a new filter:

=IF(AND(G#>=1.3,G#<1.4),"L"," ")

"L" indicates to Lay the Fav and Dutch Back the Field.

RP
Thanks for your posts. Informative. You are laying a horse, dutch laying the field and/or dutch backing the field or no bet as your options. As an integral part of this is your %WSR sheet. Hence if at say 1 minute to jump the odds are $1.30 for the fav, then your sheet says go and do "this". This is after having ascertained that you have the selection ticked off as a selection.

I have thoughts along similar lines but differ on dutching the field either laying or backing. So I am interested.

Firstly I would like to know what we can do and what we can't do and what bot we need. You are dutching at SP so the bot works out a ratio at jump -5 secs and comes up with $25 at $4 odds so you get $25 at the SP of $3.80. thus the slippage. Is there any benefit keeping the dutch bets in play. What happens to the bets still in play? Do they get matched? Do the reasonable ones get matched and the unreasonable one just expire (no match no bet no outlay)?

So what would be the outcome if say you got the ranking from the openers then put in the lay side the following 1st fav $30 @$2 lay, 2nd fav $25 @$3 lay, 3rd fav $20 @$4 etc pro rata to the openers. These are all unders. Then you also dutched the back side. Or at least just the top 90% But you did not do this at SP. You did it at say 1st fav $30 @ $4 2nd fav $25 @ $6 3rd fav $20 @ $9. Etc. These are all overs. Ultimately the figures are the %WSR plus a margin up and a margin down so you would end up with a spread. At just before jump you have an option to cancel or keep or take SP
and green up as required. Could this work and what bot could do it?

Part 6

Continue on sorting and reviewing each price subcategory. Then move on to condition breakdowns like:
Maidens only
Last 3 starts are broken up into 5 sub-categories
Weight change
Distance change
% Win Strike Rate
2nd Fav Odds
Class is broken up into 3 sub-categories
Field Size which can be very profitable

Then start combining 2 or more conditions or price sub-categories. You should end up with a heap of filters. Place them alongside one another in your spreadsheet in what I call the "Black Box".

When I see a race about to start in a minute or so's time I enter all the variables of the selection that meets my price range/favourite criteria. Once entered the "Black Box" will crunch the data and instantly generate signals "B", "L" or "N".

"N" is for No Bet. There will be times while doing your studies that you are split on the effectiveness of Backing or Laying, too close to call!
For example I have found, in the Odds/On category, that a horse "Down in Class" in any Price sub-category can win or lose on an even basis, no profit/loss to make it worthwhile. So I have a filter that generates a "N" for this situation.

I have a filter that returns 80% of liability every bet in a certain Price Range:

=IF(AND(H#<>"RFS",G#>=??,G#=3.5,M#<>"Maiden"),"L"," ")

The "Black Box" is used for staking as well. If the data entered for a selection generates say three "L" signals I increase my stake/liability by three. The skies the limit with this technique. I haven't even scratched the surface of the possibilities.

Once you become disciplined in this approach, there will be no sport that you can't find a profit in! The attachment shows a glimpse of the "Black Box" and the signals it generates. Above is a graph of the potential profit that you can achieve.
This ends this tutorial I hope you have enjoyed the insight into some of my daily activities.
RP

Beton, a reply like yours makes it all worthwhile! I know that I have limited experience and beleive me I'm happy that you've raised so many queries to perfect this system even more.

To specifics, as a rule, I seldom Dutch Lay the field if Backing the favourite. In the Odds/On category the returns make hard work of recouping losses. Backing the Favourite, as a stand alone, represents about 27% of profit.
I use BF Bot Manager Multi Strategies, Dutchbook and Multi Favourites bots, but also have Betsender and BetBotPro to look after all the different things I've got going at any one time.
Dutch Backing slippage is a real concern and I have discussed that in another thread (Races to Dutch Back). Fortunately it kind of balances out when going SP, "swings and roundabouts" if you know what I mean.
Can't comment on In Play betting, never tried it, l retired from the Building game to avoid further stress!

I hope I have spurred you and others on to entertain new ideas, it gives me a reason to get out of bed in the morning.
Cheers RP

Regards the "test" Rinconpaul.
The only things i can see are:
1) The price of the second fav is "close" to the fav, in relation to the other selections.
and 2) They are both up in distance >=400m, as are others though.

Close as I can get I'm afraid.
Keep up the good work by the way, and just because you don't get a heap of replies don't think that it doesn't get the old brain cells clinking.

How to find the edge
 

Hi RP

A very intersting thread indeed..hope I can find that edge as soon as I master my excell skills again ..been a while since I used it ..so I am still working on #2 Part 2.

Keep up the very intersting threads and helping us newbies.

Regards
Ted

Another example of filters within filters and how they can turbocharge your profits.
On the image, the greyed out section is already a filter of another group with 'maidens' and 'no change' & 'down in class' removed.

As a group it's showing a Lay the Fav, Dutch Back the field profit of $446.27 ($100 liability)

Then I have filtered it again to get two more sub groups, one to Lay, the other to Back. Can you spot what filters are used to get the two smaller groups??

Selections meeting the criteria of the top group signal a "L" and a resulting profit of $446. At the same time a second "L" or "B" signal is generated and more bets are placed resulting in another two profits, $1,259 & $381 compounding our original proft, (Beware it can also work the other way and accelerate your losses, but not as bad as when I was trading futures...lol)

In total a profit of $2,086 for a turnover of $4,700 or 44% POT

The original group, I took the greyed out selections from, was showing a breakeven scenario and that was taken from the mother sample which was showing a loss. This is how you get "blood from a stone!"

Regards the "test" Rinconpaul.
The only things i can see are:
1) The price of the second fav is "close" to the fav, in relation to the other selections.
and 2) They are both up in distance >=400m, as are others though.

Close as I can get I'm afraid.
Keep up the good work by the way, and just because you don't get a heap of replies don't think that it doesn't get the old brain cells clinking.

You'll be up for class captain if you keep this up Jose!
Yeah right on point 1. The filters were '<=6' & '>=7' of the 2nd Fav price.

Interestingly you would think a horse 'up in class' with the '2nd Fav's price' close, that the horse would Lose. Instead they Win more often? Any one explain that?
RP

Cheers mate, thought I may be missing something.
Excel still completely eludes me I'm afraid.

I've come up with an intersting stat:

81% of Odds/On favourites present in the first 4 races.

Now the first 4 races probably represent 55% of the number of races run on any day at a track?

Any theories as to why the most favoured favourites appear in the first half of a race program? Interested to hear your thoughts.....RP

The early races are usually the weakest on the card &/or contested by youngsters & can therefore be targeted by trainers with a horse that is just that bit better than the rest.

Field sizes also can be smaller early on therefore odds often skinnier.

Another gut feel at times is that punters confidence can be higher early on & they go harder than later in the day when pockets aren't so full;)

Less prize money = less runners = lower odds for the fav.