Random wagering

[Cholly]

Has anyone performed a long-term analysis of the results of truly random selection in handicapping? Previously I have casually assumed that it would produce a negative expectation that identically mirrors the take-out. But recently I’ve speculated that wagering in an absolutely random fashion would outperform (i.e., lose less) than the take-out.

The hypothesis: Assume that one combines the cumulative handicapping efforts of the entire wagering public. That combined effort will be extremely accurate, but it will never achieve perfection—there will be on some occasions a variable, not included in that joint effort, that affects the result.

The effectiveness of that joint effort will hover near 100% accuracy, but never achieve it. But…the public will wager as though their joint effort is 100% accurate, and the final odds would reflect that mistaken opinion. So a truly random wagering strategy would outperform the cumulative public, and lose less than the track take.

That’s the hypothesis; I’m sure this has been tested, but I’ve never seen the results of any such test.

[Dave Schwartz]

It will not mirror the takeout. It will be lower. Logic would indicate losing about 25%.

(The rest of this thread can be dedicated to explaining why it will be greater than takeout.)


Regards,
Dave Schwartz

[Ian Meyers]

It will not mirror the takeout. It will be lower. Logic would indicate losing about 25%.

(The rest of this thread can be dedicated to explaining why it will be greater than takeout.)


Regards,
Dave Schwartz


Dave, your estimate sounds about right to me.

One of the paradigms that will result in a HIGHER loss than takeout is named the Favorite-Longshot bias. Countless studies have demonstrated that certain types of extreme favorites are underbet while extreme longshots will win far less than public expectation.

[BillW]

When we talk takeout, don't forget to include breakage in the equation.

[Jeff P]

About a month ago I posted results of more than 100k starters in post #6 of this thread:
http://www.paceadvantage.com/forum/showthread.php?t=62616

What I was getting at in that post is that a complete newbie to the game, somebody who knows absolutely nothing about handicapping or betting - has an expectation during his or her first several trips (or even seasons) at a track that roughly mirrors picking horses at random... which for WPS betting produces long term results very much in line with the 100k starters sample that I posted...

Which IMHO is one of the many reasons racing has trouble competing with other forms of gaming... IOW current takeout levels make it all but impossible for racing to attract and keep new customers compared to other forms of gaming.

As to the reasons why picking horses at random produces less than break even less takeout less breakage... Yes, the favorite longshot bias is very much in effect.


-jp

.

[Cholly]

I’m not disputing the conclusions of Monsieurs Schwartz, Meyers or P. But I would interject that betting every horse, as per Jeff P’s data study, is not the same as randomly selecting one horse/race to bet. Maybe they yield the same result, but maybe not.

[Judge Gallivan]

I’m not disputing the conclusions of Monsieurs Schwartz, Meyers or P. But I would interject that betting every horse, as per Jeff P’s data study, is not the same as randomly selecting one horse/race to bet. Maybe they yield the same result, but maybe not.

Betting one horse randomly or betting them all you should have the same expectation and the same ROI.

Just like when playing roulette in the long run you will lose $1 per spin whether you bet $37 on 1 random number or $1 on all 37 numbers (roulette wheel with 1 zero).

[Jeff P]

Ok...

Here's the same data sample. But instead of Brisnet Prime Power, now it's broken out by RGN which is a randomly generated number between 1 and 100.
Data Window Settings:
999 Divisor Odds Cap: None

SQL: SELECT * FROM STARTERHISTORY
WHERE YEAR = 2009
AND MONTH >= 1
AND MONTH <= 6


Data Summary Win Place Show
Mutuel Totals 168396.40 167657.70 167899.10
Bet -224072.00-224072.00-224072.00
Gain -55675.60 -56414.30 -56172.90

Wins 13772 27440 40336
Plays 112036 112036 112036
PCT .1229 .2449 .3600

ROI 0.7515 0.7482 0.7493
Avg Mut 12.23 6.11 4.16
By: RGN

RGN Gain Bet Roi Wins Plays Pct Impact
5 -3090.00 10882.00 0.7160 668 5441 .1228 0.9988
10 -2396.80 11060.00 0.7833 719 5530 .1300 1.0577
15 -2990.60 11220.00 0.7335 688 5610 .1226 0.9977
20 -2225.00 10976.00 0.7973 690 5488 .1257 1.0228
25 -2172.90 11198.00 0.8060 728 5599 .1300 1.0577
30 -2841.10 11416.00 0.7511 678 5708 .1188 0.9663
35 -3320.00 11466.00 0.7104 689 5733 .1202 0.9777
40 -2378.00 11342.00 0.7903 722 5671 .1273 1.0357
45 -2148.30 11518.00 0.8135 699 5759 .1214 0.9874
50 -2906.50 10988.00 0.7355 670 5494 .1220 0.9921
55 -2522.30 11034.00 0.7714 680 5517 .1233 1.0027
60 -3127.50 11542.00 0.7290 690 5771 .1196 0.9727
65 -3392.10 11258.00 0.6987 680 5629 .1208 0.9827
70 -3393.40 10984.00 0.6911 661 5492 .1204 0.9791
75 -3398.20 11182.00 0.6961 635 5591 .1136 0.9239
80 -2653.90 11390.00 0.7670 710 5695 .1247 1.0142
85 -2505.40 11330.00 0.7789 699 5665 .1234 1.0038
90 -2854.20 11084.00 0.7425 694 5542 .1252 1.0187
95 -3060.80 13316.00 0.7701 811 6658 .1218 0.9909
Be my guest. Bet 1000's of horses at random and tell me what your expectation is.

Guarantee you that after several thousand randomly selected plays your results will mimic the above sample.

-jp

.

[Cholly]

That last batch of Jeff P's data looks pretty conclusive...how did you generate the "randomly generated number"?

[DeanT]

That last batch of Jeff P's data looks pretty conclusive...how did you generate the "randomly generated number"?
Jeff is sneaky, you have to watch him...... his favorite number is 25. :liar:

[Jeff P]

It's built into JCapper. The compiler that I used has a built in random number generator. It's not a true or perfect random number but it's close.

RGN actually has a purpose besides being able to sneak it into data samples posted on message boards every now and then. I use it to segregate data samples.

I keep waiting and hoping to see a 25 horse field... Short of that I guess I'll have to settle for a 2-5 exacta in a 10 horse field. ;)

-jp

.