Keeping records for success at different odds, what to use?

[cutchemist42]

After reading a few books lately, I'm kind of starting to think I should see what my success rate truly is at different odds levels. Just kinda wondering though as a Win/DD/P3 better only what would be useful to me? Should I do it based on my odds line I set, or the ML, or the closing line? When doing DD/P3s, you can sometimes only judge how well the ML is so how should I handle single keys in DD/P3s?

Thanks for any help/suggestions from anyone that tracks this themselves.

[DeltaLover]

Originally Posted by cutchemist42

After reading a few books lately, I'm kind of starting to think I should see what my success rate truly is at different odds levels. Just kinda wondering though as a Win/DD/P3 better only what would be useful to me? Should I do it based on my odds line I set, or the ML, or the closing line? When doing DD/P3s, you can sometimes only judge how well the ML is so how should I handle single keys in DD/P3s?

Thanks for any help/suggestions from anyone that tracks this themselves.

I keep it as simple as possible, recording PNL in a per race basis keeping a very short note about the most significant horses ( favourites that I go against or the horses I decided to bet), morning-line of final odds are not part of my record keeping (except of course for PNL reasons)

[HUSKER55]

If you wager with Twin-spires they have a section where they record your betting stats.

[JJMartin]

I would say look at your odds line and see how well you do against the m/l especially on multiple leg bets.

[Ray2000]

If you're familiar with Excel filtering and charting you can collect your results on a spreadsheet, sort the wagers on the odds column and plot the running balance as if they were bet in the sorted order. This will give you a method to 'eyeball' the positive slopes where profits can be had, if any. This sorting can be done on any of the categories where you collect data,... Morning Line, Speed, Class, Turf etc., to find where your best results lie.

As a template to demonstrate this, I've attached the spreadsheet I use*. It has the data from my last test run here at PA. You can delete that after getting a feel for using the spreadsheet.

Sheet "Data" will contain all the raw data on the wagers you've made. (If you change any column heading names here you'll need to use the same names on sheet "Selects").

Sheet "Selects" has a macro that will filter out races based on the criteria in cells A1:I5 and place the requested data in rows A23:I32000.
Columns J,K,L have the formulas for balance and the X label index. The X axis is dynamic and will change based on the number of races selected.

The chart can plot 3 different series, all based on $2 returns. I have "Place" plotted twice (one on top of the other) because no data is available for Exacta or Trifecta in this example. You can change the names in G22, H22, or I22 to fit your specific result column names.



*Forum won't let me attach xlsm file so I put it at:

http://ge.tt/2YQphI12/v/0?c


Any questions, I'll try to answer.

[Robert Goren]

Originally Posted by cutchemist42

After reading a few books lately, I'm kind of starting to think I should see what my success rate truly is at different odds levels. Just kinda wondering though as a Win/DD/P3 better only what would be useful to me? Should I do it based on my odds line I set, or the ML, or the closing line? When doing DD/P3s, you can sometimes only judge how well the ML is so how should I handle single keys in DD/P3s?

Thanks for any help/suggestions from anyone that tracks this themselves.

Why not track them all? It just another line or two on an excel spreadsheet. Then you can sort them into any and all groups you want to.

[MJC922]

IMO the ‘fly in the ointment’ when it comes to handicapping research is the notion that we can actually obtain the prices we’ve seen on paper when it comes to real world execution.

For example here’s a recent method I’m tracking:

Plays 23825
Wins 7934
Wpct 33.3%
Invested $47,650.00
Returned $46,166.80
Profit / Loss -$,1483.20
ROI -3.1%

First some background on the above, this was a method developed using one month of results from every track in North America and then forward tested on four months to mitigate against back-fitting. There’s no minimum odds requirement shown above so this is more or less a $2 win bet on every race for a 5 month period. As you can see the selection essentially wins as often as the public choice and loses 3% on the dollar.

Now let’s add a minimum odds requirement of 5-1:

Plays 2654
Wins 381
Wpct 14.4%
Invested $5,308.00
Returned $6,391.40
Profit / Loss +$1,083.40
ROI +20.4%

Ok so now we have something ‘potentially’ anyway. In a month-to-month breakdown, the worst of the five months saw this method get a +10% ROI with the odds requirement enforced. Now this was all on paper.

Let’s look at the challenges presented by real world play… the following is snaphot detailing the past 8 days of action. I’m using Twinspires conditional wagering feature with minimum odds of 5-1 at 0 MTP:

Plays 97
Wins 14
Wpct 14.4%
Invested $194.00
Returned $176.20
Profit / Loss -$17.80
ROI -9.1%

Digging deeper, here are the payoffs on the 14 winners:

19.60
17.40
16.80
16.20
13.80
13.60
13.20
11.80
11.00
10.40
10.20
9.60
8.20
4.40

What immediately jumps out as a red flag at this early stage half of the winners paid less than the minimum odds needed to maintain the target ROI. I thought the $4.40 horse was worth investigating as a possible mistake by the ADW but sure enough they were all loaded into the gate and he was 9-2, during the running of the race you could see the odds drop several times until he was 6-5 in mid-stretch. Granted at larger tracks these types of radical changes are going to be less severe but the issue remains.

Conclusions: You see ‘on paper’ play, you’re always getting your price. Some horse’s that are 5-1 in the result chart were probably 3-1 at 0 MTP, and on paper we get all of that gravy. Worse is the huge hit we're taking on price, i.e. a staggering percentage of what should be 5-1 (or more) turns out to be nothing but a mirage in the end. Half of the winning horses wouldn’t have even been considered ‘plays’ on paper.

I'll keep others posted on these results moving forward but given what I've seen so far it wouldn’t surprise me at all if this ‘method’ loses in real-world play. One of the biggest problems in the modern game is the very basis of profitable play (evaluating risk vs. reward) has been undermined and we’re all too often flying blind on assessing the reward-side. For lack of a better term that would seem to be a suckers game.

[ReplayRandall]

Is your method for thoroughbreds?

[Ray2000]

Originally Posted by MJC922

....minimum odds of 5-1 at 0 MTP

You might want to try an ADW that allows..... -2 MTP

Higher risk of shutouts, but I've found a big difference that way in ROI

[lansdale]

Originally Posted by MJC922

IMO the ‘fly in the ointment’ when it comes to handicapping research is the notion that we can actually obtain the prices we’ve seen on paper when it comes to real world execution.

For example here’s a recent method I’m tracking:

Plays 23825
Wins 7934
Wpct 33.3%
Invested $47,650.00
Returned $46,166.80
Profit / Loss -$,1483.20
ROI -3.1%

First some background on the above, this was a method developed using one month of results from every track in North America and then forward tested on four months to mitigate against back-fitting. There’s no minimum odds requirement shown above so this is more or less a $2 win bet on every race for a 5 month period. As you can see the selection essentially wins as often as the public choice and loses 3% on the dollar.

Now let’s add a minimum odds requirement of 5-1:

Plays 2654
Wins 381
Wpct 14.4%
Invested $5,308.00
Returned $6,391.40
Profit / Loss +$1,083.40
ROI +20.4%

Ok so now we have something ‘potentially’ anyway. In a month-to-month breakdown, the worst of the five months saw this method get a +10% ROI with the odds requirement enforced. Now this was all on paper.

Let’s look at the challenges presented by real world play… the following is snaphot detailing the past 8 days of action. I’m using Twinspires conditional wagering feature with minimum odds of 5-1 at 0 MTP:

Plays 97
Wins 14
Wpct 14.4%
Invested $194.00
Returned $176.20
Profit / Loss -$17.80
ROI -9.1%

Digging deeper, here are the payoffs on the 14 winners:

19.60
17.40
16.80
16.20
13.80
13.60
13.20
11.80
11.00
10.40
10.20
9.60
8.20
4.40

What immediately jumps out as a red flag at this early stage half of the winners paid less than the minimum odds needed to maintain the target ROI. I thought the $4.40 horse was worth investigating as a possible mistake by the ADW but sure enough they were all loaded into the gate and he was 9-2, during the running of the race you could see the odds drop several times until he was 6-5 in mid-stretch. Granted at larger tracks these types of radical changes are going to be less severe but the issue remains.

Conclusions: You see ‘on paper’ play, you’re always getting your price. Some horse’s that are 5-1 in the result chart were probably 3-1 at 0 MTP, and on paper we get all of that gravy. Worse is the huge hit we're taking on price, i.e. a staggering percentage of what should be 5-1 (or more) turns out to be nothing but a mirage in the end. Half of the winning horses wouldn’t have even been considered ‘plays’ on paper.

I'll keep others posted on these results moving forward but given what I've seen so far it wouldn’t surprise me at all if this ‘method’ loses in real-world play. One of the biggest problems in the modern game is the very basis of profitable play (evaluating risk vs. reward) has been undermined and we’re all too often flying blind on assessing the reward-side. For lack of a better term that would seem to be a suckers game.

Hi MJ,

After showing results from much larger samples, you base your negative conclusion on a 100-odd race sample? Why? It's meaningless.

Also, BTW, the first, largest sample results you posted would be profitable with a rebate - +7.1% with a 10%, and profitable even with a 5% rebate. And with a 10% rebate, your second, smaller sample would have a 30.4% edge. With a professional $200 bet rather than a $2 bet, this would yield ca. $180,000 for about 2 years of full time play. I believe this is the first time I've ever heard anyone whine about having a successful model.

BTW, I'm not arguing with the premise that the odds-drop doesn't hurt the average bettor, but you're obviously far from being the average bettor.

Cheers,

lansdale

[MJC922]

Originally Posted by lansdale

Hi MJ,

After showing results from much larger samples, you base your negative conclusion on a 100-odd race sample? Why? It's meaningless.

Also, BTW, the first, largest sample results you posted would be profitable with a rebate - +7.1% with a 10%, and profitable even with a 5% rebate. And with a 10% rebate, your second, smaller sample would have a 30.4% edge. With a professional $200 bet rather than a $2 bet, this would yield ca. $180,000 for about 2 years of full time play. I believe this is the first time I've ever heard anyone whine about having a successful model.

BTW, I'm not arguing with the premise that the odds-drop doesn't hurt the average bettor, but you're obviously far from being the average bettor.

Cheers,

lansdale

Hi there. Thanks for the reply. I don't disagree, a 100 race sample is meaningless and half of the winners coming in below the cutoff is once again rather meaningless as I suspect that's probably higher than it will ultimately turn out to be. With that said, having no earthly idea what you're going to get paid half the time is a bit like shopping at a supermarket where the price tags are clearly posted on the shelf but a good portion of the items in your cart ring up 2 to 3 times higher at the checkout. Probably not a wise place to keep coming back to. Leaving rebates aside for now -- whether the model ends up winning or not, it's pretty clear it will be nowhere near 20% ROI and probably nowhere near 10% ROI in real world play. The post is more about the people looking at their own records and saying you know if I could just wait for the right prices I would be a winner, and I would have to say it's not quite as simple as that. Whatever arbitrary cutoff we make based upon those recorded final odds there will be no practical way to enforce it.

[MJC922]

Originally Posted by ReplayRandall

Is your method for thoroughbreds?

Yes, exclusively.

[MJC922]

Originally Posted by Ray2000

You might want to try an ADW that allows..... -2 MTP

Higher risk of shutouts, but I've found a big difference that way in ROI

Haha thanks for the tip. I may have to go that route, although the reduction in plays must be pretty sharp. I'd hate to have to get to the point where the only plays that I make are when one of the horses is a gate scratch.

[lansdale]

Originally Posted by MJC922

Hi there. Thanks for the reply. I don't disagree, a 100 race sample is meaningless and half of the winners coming in below the cutoff is once again rather meaningless as I suspect that's probably higher than it will ultimately turn out to be. With that said, having no earthly idea what you're going to get paid half the time is a bit like shopping at a supermarket where the price tags are clearly posted on the shelf but a good portion of the items in your cart ring up 2 to 3 times higher at the checkout. Probably not a wise place to keep coming back to. Leaving rebates aside for now -- whether the model ends up winning or not, it's pretty clear it will be nowhere near 20% ROI and probably nowhere near 10% ROI in real world play. The post is more about the people looking at their own records and saying you know if I could just wait for the right prices I would be a winner, and I would have to say it's not quite as simple as that. Whatever arbitrary cutoff we make based upon those recorded final odds there will be no practical way to enforce it.

You must realize that the supermarket analogy doesn't really work unless you're placing bets with the cashiers. There will be variance. You seem not to grasp that the first model above, let's call this Model #1 had an ROI of .97 *with no odds filter*. That's a pretty strong model - one which should easily be able to be tweaked for a positive ROI, and as I said, is profitable with a rebate (why would you dismiss this?).

Your concern about the odds-drop is overblown. I'm assuming that your second sample above is extracted from your first larger sample and that your results for both are based on final odds. If so, the only late-odds movement that should make you question your results is on horses whose odds would have risen after closing, to the target of 5-1 or higher, and won. Their number should be miniscule and any loss thereby should be more than made up by the dropping-odds winners you're complaining about. The robustness of your model is far more important than the odds-drop.

I would wait until you get to at least 1k races with this to start questioning your model. If you continue to post going forward I think it could be helpful for everyone.

Cheers,

lansdale

[MJC922]

Originally Posted by lansdale

I would wait until you get to at least 1k races with this to start questioning your model. If you continue to post going forward I think it could be helpful for everyone.

Cheers,

lansdale

I'll provide a status on this from time to time. It had a good week, looks like it finally caught a couple of decent prices and got into the black, the totals after two weeks now look as follows:

# of Bets Win % $1 ROI Wagered Payoff Profit/Loss
162 14.81% +0.03 $324.00 $332.90 $+8.90

10 of the 25 (40%) winners went off below the requested 5-1 conditional minimum odds.