Consecutive Loss Estimates

[InControlX]

I think I share a character trait with most handicappers that losing streaks tend to demoralize me and lead me to doubt methods and my dedication to the task. One possible corrective measure is to devise "insurance bets" which while not yielding the ROI of primary wagers at least lessen the drought periods between ticket cashing and keep the stress down.

Towards this aim I began investigating what sort of consecutive losses I should expect based upon a winning wager percentage and number of plays. To get these results, I waded through trend probabability statistics and verified them with a simple Visual Basic program. Here's the results for a sequence of 1000 plays (typical annual for me) in most probable consecutive loss counts rounded to integers over a one-hundred run average:

..................Consecutive Losses
..........5 6 7 8 9 10 11 12 13 14 15 16 17
Win %
30% ...28 21 15 10 7 5 4 3 2 1 1 1 0
35% ...28 18 12 8 5 3 2 1 1 0 0 0 0
40% ...24 16 9 5 4 2 1 1 0 0 0 0 0
45% ...20 12 6 3 2 1 1 0 0 0 0 0 0

For example, if your handicapping yields 30% winners and you place a series of 1000 bets, the most probable consecutive loss results will be 5 losses in a row 28 times, 6 losses in a row 21 times, 7 losses in a row 15 times etc.

Perhaps most startling are the longest losing streaks, even reaching 11 in a row for a 45% winning sequence! Also, the idea that I'm likely 20 times to lose five in row over the year is a mind bender. This certainly seems to favor insurance bets.

ICX

[podonne]

Making insurance bets may be more or less helpful depending on the ROI of your primary wagers. By definition, insurance correlates return and risk, so if your purpose is to lower risk, then you neet to have enough "cushion" in your ROI to account for the loss due to your insurance bets.

[InControlX]

Podonne,

Let's just assume you have an overall good ROI on the primary bets, say 150% over the year. The ROI does not change the consecutive loss probabilities. You will naturally lower your overall ROI by making the less-profitable insurance bets, but you carve up the losing streaks and retain your hair.

ICX

[cmoore]

Losing streaks can make you do many things.

Bet more $$$ then usual trying to get losses back
Bet many more races
Bet very low priced runners that in the long run are losing bets
Bet high priced horses hoping for a big hit without any legitimate reasons why

Sometimes just cashing a ticket for a small profit can shake that losing streak..If you can figure something out that minimizes long losing streaks and that helps keep yourself sane. Go for it.

[LottaKash]

Losing streaks can make you do many things.
..If you can figure something out that minimizes long losing streaks and that helps keep yourself sane. Go for it.

Yes, as a retired person, on a fixed income, my money is always "scared money", that is why I always make a place wager along with my other wagers....I am like the "energizer bunny", I just keep going and going and going...:jump:

best,

[DanG]

Interesting topic InControlX; :ThmbUp:

If any excel user is interested in a neat formula to “simulate” the classic gamblers ruin formula; here’s a post I ran on the HTR bbs a while back. In the interest of time saving I’ll just copy and paste it. (BTW: I’m 9 miles from teaching a computer class so please forgive the generic nature of this.) Original thread: http://www.homebased2.com/forums/showthread.php?t=8036&highlight=excel

========================================

Assuming you have Excel here is a handy formula for infamous “gamblers ruin” losing streak formula.

It involves two nested functions:

1. Round:
2. LN: Returns the natural logarithm of a number.

You need to fill in 2 cells:

1. Sample Size (A1)
2. Hit rate (B1)

For this example let’s say your using cell “A1” for sample size and “B1” for hit rate. Sample of 100 and hit rate of 20% which is entered as .20 not as an integer.

Formula;

• =(ROUND(LN(A1)/-LN((1-B1)),0))

Literally…=(ROUND(LN(100)/-LN((1-.20)),0))

Reminder that “LN” is a function not a cell address.

If you don’t have Excel, here’s a chart of theoretical losing streaks based on a sample size of 100 and various win percentages.
HIT-RATE PROJ LOSING STREAK
01% 458
02% 228
03% 151
04% 113
05% 090
06% 074
07% 063
08% 055
09% 049
10% 044
11% 040
12% 036
13% 033
14% 031
15% 028
16% 026
17% 025
18% 023
19% 022
20% 021
21% 020
22% 019
23% 018
24% 017
25% 016
26% 015
27% 015
28% 014
29% 013
30% 013
31% 012
32% 012
33% 011
34% 011
35% 011
36% 010
37% 010
38% 010
39% 009
40% 009
41% 009
42% 008
43% 008
44% 008
45% 008
46% 007
47% 007
48% 007
49% 007
50% 007

In closing…

If you’re playing that spot play who hits at 2% and on your 228 straight loss...bless you my son and may your pockets be deep and your blood pressure remain stable! :jump:

[timtam]

I was always under the impression that if you had a large enough bankroll

you could make money just playing the favorite because of its common

win percentage. I know about small payoffs on odds-on but just winning one

of those will keep you in the hunt until you get the $6-$7 favorite payoffs.

I'm not saying huge progession wagers just a small increment up after every

time you lose. I have an old book called Scientific Wagering Systems and

you start with $5 bet the your increments are up $1 until you may wager

$12 and thats where you stop making $12 bets sliding down maybe one or

two slots after a hit. This is an old book where the guy makes $12 for 2

days of play which in this day and age is unrealistic but maybe with enough

of a bankroll you could turn that $12 into $120. The favorites win above

30% at most tracks and if playing a favorite system you can't NOT BET

the odds-on. It just doesn't make any sense.

[CincyHorseplayer]

Podonne,

Let's just assume you have an overall good ROI on the primary bets, say 150% over the year. The ROI does not change the consecutive loss probabilities. You will naturally lower your overall ROI by making the less-profitable insurance bets, but you carve up the losing streaks and retain your hair.

ICX

I have wrestled with this too.If I were Barry Meadow I would chastise and demoralize you all day for the concept but I agree that bets are a totality,an investment whole,and it has to be looked at that way.These insurance bets do lessen the blow and they soothe the psyche.Let's say for instance;

Your avg win mutuel=10.00/place=4.00

And your win rate is 30%,place rate 60%.

Scenario A- you bet $4 to win,every 10 bets you get a 50% ROI

Scenario B- you bet $2 w/p,every 10 bets you get a 35% ROI

You lose 15% ROI but if you are on a losing streak,in scenario A its a 100% loss,in scenario B you cut losses to 40%.

More importantly you stay alive longer.Theoretically if you started with $40(making $2 wp bets) and you lost every win bet during the streak but maintained your place hit rate at 60% you would double the amount of bets you can place than if you were just win betting alone.Your bankroll would drop from $40-24-14.40-7.60.or 10 bets to 6 to 3 to 1=20 bets(21 actually)

I think knocking off some of the ROI off the top to double your lifespan is worth doing.

I bet exactas so I don't figure it out by average mutuels but by average payout and separate them into 2 categories=primary and secondary,then apply the hit percentages,put them into an array and come up with composite ROI's,for instance here are my composite ROI's from my projection method;

18% hit rate=37% ROI
21% hit rate=50% ROI
25% hit rate=63% ROI
29% hit rate=87% ROI
32% hit rate=110% ROI
36% hit rate=123% ROI

If you play exotics just separate higher and lower payoffs and come up with averages,then take your hit percentages and stagger the numbers up and down each column(say 8 hits go 1-7/2-6/3-5/4-4 etc) to come up with average payouts,add them together and divide by the number of payouts to come up with a composite ROI for each % of the win bracket.Everybody's is different.

[lamboguy]

we are dealing on the face of things at a zero sum game. by the math no one is supposed to win at horserace betting. when you break it down, 10% takeout in the wps pools are impossible to overcome by just trying to pick winners. i used that number because if you are a rebate player that is generally how much of a percentage you pay no matter which track you play.

there are a few ways to put the odds in your favor. first you must know ahead of time that the horse you are playing is as sound as sound can be before the race. if unsound don't play the race. try to find out if there was anything done to the horse inbetween starts that would improve his performance. the other thing i would do is go to the racetrack every morning, sit with the official racing form clockers and watch the horses work. learn to diferentiate a good work from a bad one.

[dav4463]

I was always under the impression that if you had a large enough bankroll

you could make money just playing the favorite because of its common

win percentage. I know about small payoffs on odds-on but just winning one

of those will keep you in the hunt until you get the $6-$7 favorite payoffs.

I'm not saying huge progession wagers just a small increment up after every

time you lose. I have an old book called Scientific Wagering Systems and

you start with $5 bet the your increments are up $1 until you may wager

$12 and thats where you stop making $12 bets sliding down maybe one or

two slots after a hit. This is an old book where the guy makes $12 for 2

days of play which in this day and age is unrealistic but maybe with enough

of a bankroll you could turn that $12 into $120. The favorites win above

30% at most tracks and if playing a favorite system you can't NOT BET

the odds-on. It just doesn't make any sense.


I would say to try this, but only play favorites that you like. You should be able to eliminate more than half the favorites on a card and just play the ones you think should win. If you can just increase your win percent a little bit, then it might work. You also have to avoid the temptation of betting your pick when it is not the favorite. If it is a favorite system, then stick with favorites only.

[maxwell]

I found DanG's formula to be pretty darn close to my records of the last five years or so. But not every year, which makes sense due to "chance", which is the deviation in probability. I think "insurance bets" a.k.a. "hedge plays" are an important part of the game. They can save your year and /or pride in one of the world's toughest games.

Interesting thread! :)

[InControlX]

DanG,

Thanks for the input. Your numbers agree with the ones I found in the statistical testing books (curiously for medical trials) and confirmed using the VB random function on averaged runs. It's cool that there's a log function to predict this.

Maybe now knowing that these consecutive losses are normal and to be mathematically expected as opposed to due to some outrageous fortune will help me suffer through the droughts and refine the insurance wager. Thanks for the insurance strategy suggestions from others.

I'm going to run some different insurance bet strategies on my data base and see if some help to bridge the winning gaps. I'll try some place, show, and exactas under different mixes with 2nd picks.

ICX

[Niko]

The formula is pretty slick, thanks Dan G.

I was looking for something like this. Is there a formula that tells you what your max drawdown would be for flat bets over a range of bets?

For example; you could lose 15 in a row, win 1 and then lose another 10 in a row. Maybe it's just a function of looking at the win percentageand then the expected losing streaks to figure it out. Similar to what InControlX must have written.

I've seen charts for 50% drawdowns using a percentage of Kelly, but nothing with flat bets.

I've been out of school for a long time so I need to brush up on math. I hate to learn it all over again but maybe that's what I'll have to do.

[BlueShoe]

At a Hollypark meeting back in the 60's post time favorites lost 34 times in a row on one occasion.My own personal worst losing streak was 26 losers in a row some years ago during a very bad Santa Anita meet.These were mixed priced horses,with most in the mid price range.The moral to the story is:It can get a lot worse than you think.Even worse than the rare long losing streak is the long bogged down period,where you are hitting an occasional winner,but far below your normal expectation.

[Robert Fischer]

http://www.paceadvantage.com/forum/showthread.php?p=320538&highlight=consecutive#post320538

[Niko]

thanks robert;

I was looking for not only the probablitiy of consecutive losses but also if there a way to tell the probabilities of the max drawdown-or a bad losing streak over time.

In other words; not consecutive losses but a prolonged losing streak where you could have some wins in the sequence.

I'll have to check some of the futures programs that have monte carlo simulations etc, but I think those go by historical data. Back to google when I get a chance

[CincyHorseplayer]

thanks robert;

I was looking for not only the probablitiy of consecutive losses but also if there a way to tell the probabilities of the max drawdown-or a bad losing streak over time.

In other words; not consecutive losses but a prolonged losing streak where you could have some wins in the sequence.

I'll have to check some of the futures programs that have monte carlo simulations etc, but I think those go by historical data. Back to google when I get a chance

This is part of what I was getting at in my previous post.Over time,after keeping records for 7 years,I was able to come up with an accurate projection method based on number of races played,the actual number of hits divided into percentages,and the average payoff(high and low).Then you just start from low to high in percentage brackets to get an estimate of how it will affect your bankroll over a week,2 weeks,etc.For example;

Say you start with a $1200.00 bankroll and you bet 2%($24) of it 4 times a day,5 days a week for a weekly average of $480.00 a week on 20 bets.

We'll just say your average payoff is $100.00

5% =100.00-480.00=(-380.00)-1 hit(79% loss)
10%=200.00-480.00=(-280.00)-2 hits(58% loss)
15%=300.00-480.00=(-180.00)-3 hits(38% loss)
20%=400.00-480.00=(-80.00)-4 hits(17% loss)
25%=500.00-480.00=(+20.00)-5 hits(4% ROI)
30%=600.00-480.00=(+120.00)-6 hits(25% ROI)
35%=700.00-480.00=(+220.00)-7 hits(46% ROI)
40%=800.00-480.00=(+320.00)-8 hits(67% ROI)
45%=900.00-480.00=(+420.00)-9 hits=(88% ROI)
50%=1000.00-480.00=(+520.00)-10 hits(108% ROI)

You can stretch the numbers by percentile brackets over any stretch of bets or time period you want once you come up with the average number of bets you make,the % of bankroll of those bets,and your average payoffs.Remember you should always project out winning AND losing %.Not just expected losing brackets over time or loss streak %.It's good to know what the bottomline will be of losing runs we all experience but it's excellant to know what kind of damage you can do when you are locked in.