Flat Kelly staking

[skylark2009]

Flat Kelly Staking.

 

1. Start bank = 100pts

2. Flat Kelly stakes used as opposed to full Kelly stakes.

Full Kelly stakes are usually derived as a percentage of the current bank, increasing/decreasing in size depending on the suggested Kelly percentage of the current size of the bank. This makes full Kelly staking a very progressive staking method and high risk (especially so if your assessments prove to be too high and loses).

I'll unlink from the bank by converting the suggested Kelly percentage straight to points (essentially locking the bank at 100pts).

• Derived Kelly stake suggests 14.59% of the current bank, then the stake will be 14.59pts
• Derived Kelly Stake suggests 7.43% of the current bank, then the stake will be 7.43pts etc

This should flatten the stakes and make it a lot less progressive (hopefully).

 

@BillyHills

Hope this is ok to post here, if not then move to Glory Hunters.

 

[skylark2009]

Bet#1  2.10 Lingfield - Meringue @1/1 - 13.04pts

 

Edited September 5, 2018 by skylark2009
Added stake

[Pirate53]

2.1 available at Matchbook. (0% commission on horses).

[skylark2009]

Bet#2 Ffos Las 6.30 Spirit Ridge @ 5/4 - 13.52pts

[Sir Puntalot]

@skylark2009 The problem with this type of thread being in ATR is that others start to do the same and then the daily chat threads lose out.

I'll let @BillyHills decide, but I think it needs to be in S&S or GH myself.

Either way, good luck with it.  

[BillyHills]

I don't mind it in here to be honest, we already have one or two others posting individual threads.
If it gets silly we can always have a rethink.

Also i like to see the racing section busy with different types of threads and it may encourage others to start threads or add posts. We need more people posting thats for sure!

 

[Sir Puntalot]

Consider me slapped down!  

Carry on @skylark2009 - I'll get rid of these posts out of your thread once you've read it.  

[skylark2009]

@Sir Puntalot

No problem with any feedback, so the posts are fine to stay where they are.

Also, just to vary this thread so its simply not just a straight up 'selection and P/L' thread, I'll include my thoughts on staking strategies etc, which may help generate responses here.

 

[skylark2009]

Yesterdays bets

 

#1. Meringue - An R4 of 10p meant a reduction in returns from the win.

#2. Spirit Ridge - lost (had I waited closer to the racetime, then the stake would have been greatly reduced )

 

Totals to date -

 

 

Edited September 6, 2018 by skylark2009

[skylark2009]

Kelly criterion. (taken from various sources)

The Kelly criterion was devised by John Larry Kelly jr in 1956 as a means to maximise long term growth from investments. It has subsequently been used as a stake generating method in various forms of betting (football, horse racing etc), where the bettor thinks he has an edge.

Nb. if your edge is equal to or less than the current odds available, then a 'No Bet' situation would occur.

The formula is -

• % Stake  = ((Decimal Odds x % chance of winning) - 1) / (Decimal Odds  - 1) x 100

Where

• Stake = % of the current bank  to use
• Decimal odds = available odds as decimal (if using fractional odds then convert to decimal)
• % chance of winning = your assessed chance of success (your edge)

 

E.g. (using yesterdays two bets)

1. Meringue -

• Odds available = 1/1 (decimal odds of 2)
• my assessed chance of success = 56.52% (0.5652)

putting these into the formula is -

• ((2 x 0.5652) - 1) / (2 - 1) x 100  = 13.04% of the current bank.

2. Spirit Ridge -

• Odds available = 5/4 (decimal odds of 2.25)
• My assessed chance of success = 51.95% (.5195)

putting these into the formula is -

• ((2.25 x 0.5195) - 1) / (2.25 - 1) x 100  = 13.52% of the current bank.

Pro's/Cons

+ Probably an efficient way to grow a bank (if the edge assessment is also correct)

- Is progressive in nature
- Is high risk (especially in the case of over assessing the chance of success, which then loses)
- Requires a good working background of assessing/calculating percentages (your edge)
- It will not turn a losing selection process into a profitable one
- Is able to generate very high stakes (in relation to the bank) given the variables that are input

[skylark2009]

2.50 Newcastle - Trouble And Strife (to place, 2 places) - 3.94pts @ 1/2 SkyBet

 

A Sir Mark Prescott profile for this one -

• 3 or 4 runs as a 2yo without winning.

 SMP has an excellent record for doing this with his 2yo's when racing as 3yo's, usually running them over further too. Because they haven't won as a 2yo, they would also tend to have low OR's as a result.

Trouble And Strife hasn't been out of the first two places since resuming racing as a 3yo (six runs) and also managed two wins to boot.

 

My assessed chance of at least placing is 67.98%, coupled with an available price of 1/2 to place (two places). Using the Kelly formula means a flattened stake of 3.94pts is used for the bet. 

[skylark2009]

#4 2.05 Kempton Enable @ 8/13 12.51pts

 

Last year's Arc winner drops in class after a long layoff. If race ready, she could continue onwards here. She's suffered only one defeat and is in receipt of 8lb's from the in form Crystal Ocean.

 

Assessed chance here = 66.67% with odds of 8/13 = Flat Kelly stake of 12.51pts 

 

[skylark2009]

#5 4.25 Kempton Quiet Endeavour @ 5/4 6.0pts

 

Seems to be on a roll with four recent wins and a course/distance winner to boot.

 

Assessed chance of winning 47.78% @ odds of 5/4 = Flat Kelly stake of 6pts

[skylark2009]

#6 2.55  York Limato @ 8/13 - 24.96pts

1. Kimbrella - He has contested in 9 listed races, 3 group 3's and a group 1 in the last three years managing to win two of those listed races. Won twice here (8 races total), but does seem to prefer a sharper track. He is rated 9lbs lower than Limato but carrying the same weight as him today.

2. Limato - He has contested 17 class 1 races in total, winning eight of them (including two group 1's) .  Top rated here and is 4lbs better than Stormbringer (rated 105) at the weights. Doesn't seem to be the force of old but is very capable in his day.

3. Justanotherbottle - High rated handicapper trying listed  company for the first time. Form line through Gifted Master suggests he has something to find here with Limato.

4. Tropics - 18 class 1 races in total, winning four of them (including a group 3). He has been running mainly in class 2/3 recently and will likely need to step up here.

5. Stormbringer - Won a class 5 and has been progressing up the ratings with some good efforts,  being tested against the older horses here for the first time, big improvement required here as I can't see his WFA allpwance really helping him here.

Limato looks the class horse here, some form lines through Gifted Master/Spring Loaded suggest that Justanotherbottle and Kimbrella have something to find, while Tropics and Stormbringer need to equal or improve on their personal bests.

8/13 (61.9%) for Limato, while it looks short is probably about right  here imo. I would actually rate his chances here closer to 2/5 (71.41%).

My assessed chance of winning, 71.41% against an available price of 8/13 = Flat Kelly stake of 24.96pts

[skylark2009]

Last 6 bets -

 

 

 

[MCLARKE]

Always a bit wary about staking plans, I increase my stake by £1 every time my bank has increased by a certain amount, it won't make me a millionaire but has meant that my average weekly profits have been increasing over time. It also means I don't bet more than I feel comfortable with.

Out of interest how do you calculate your assessed chance of winning?

[skylark2009]

33 minutes ago, MCLARKE said:

Always a bit wary about staking plans, I increase my stake by £1 every time my bank has increased by a certain amount, it won't make me a millionaire but has meant that my average weekly profits have been increasing over time. It also means I don't bet more than I feel comfortable with.

Out of interest how do you calculate your assessed chance of winning?

@MCLARKE

Agree here, very sensible to keep a good control over your stakes and not let previous results affect future staking until you decide to. Level stakes with one off progressions when my bank hits a certain level is something I use frequently.

 

My assessments are mainly derived from an Excel spreadsheet I started in 2001, which contains a data list containing all flat winning horses (UK) from 1998. I then group these horses in various ways when they run again.

These groupings have winners/losers, and after a little further filtering of each group I can determine a winner/runner ratio for that group.

 

E.g. Last time out flat winners running today (10/9/18).

 

Column T6 are the unfiltered groupings.

Odds1 shows no. of runners in that group with an assessed price. e.g Solveig's Song falls into a group that has had 702 runners and winner/runner ratio of 9/2 or an 18.18% SR. Further filtering on this group allows me to increase/decrease this ratio to determine a potential edge.

The 2nd Odds column are the current prices.

[skylark2009]

#7 2.99 Brighton Arcanista to place (2 places) @ 4/5 Skybet - 8.07pts

Arcanista won last time over course/distance, she could could be capable of maintaining her form again today. The S De Sousa jockey factor (S De Sousa in small fields) could well help here.

Assessed probability to place is 59.23% against odds of 4/5 (2places) with Skybet = Flat Kelly stake of 8.07pts used.

[MCLARKE]

Thanks for the reply. Looks like an interesting method, might have a look at doing something similar myself. Are your selections based purely on the calculations or are they influenced by other subjective factors, e.g. is the S De Sousa factor part of the calculation?

[skylark2009]

@MCLARKE

 

The selections are mostly from the spreadsheet (probably 95%+), but I do use other means (Proform, Horseracebase and reading form etc) for some of my bets. If anything I put up here won its last race, then it will likely be from the spreadsheet and wouldn't be influenced by other factors.

 

---------------------------------------------------------------------------------------------------------------------------------------------------------------------------

#8 Chelmsford 6.15 Fognini to win @ 5/4 Bet365 - 8.12pts

Four runs with two wins. Fognini has the most experience of the field and has to give weight to his rivals here, but may be able to claim another win here.He beat Billy No Mates (6th) last time, who has won since, but five others from this race have also run again, each failing to further frank this form. Invincible One looks next best here reading through the form of his last race, which has produced two next time winners.

Assessed odds of 48.95% against an available price of 5/4 = Flat Kelly stake 0f 8.12pts

 

 

[skylark2009]

#9. 4.35 Salisbury Jash to win @ 2/5 SkyBet - 17.15pts

Won his maiden last month with what seemed like something to spare. The 3rd, 4th and 5th in that race have since run again. Riviera Nights (3rd) won next time, Jack's Point (4th) won next time and Kamikaze Lord (5th) managed 2nd next time in what looked like a competive race.

My assessed win chance is 76.33% against odds of 2/5 for a Flat Kelly stake of 17.15pts

 

#10. 7.10 Salisbury Altra Vita to win @ 15/8 William Hill - 11.52pts

In fine form this year with four wins. Sir M Prescott runner, who as a 2yo failed to win (in three runs). SMP has a knack for doing this when running them at 3yo, often beating the handicapper. Altra Vita needs to run from the front here for this as Encryption is a very credible danger here.

Assessed chance of winning is 42.49% against odds of 15/8 = Flat Kelly stake of 11.52pts

[skylark2009]

Yesterdays two bets were non runners due to Salisbury being called off. Today's bet becomes bet #8.

 

#8. 7.45 Kempton - Twister to place (2 places) @ 4/6 SkyBet 8.78pts

Drifting compared yesterday's opening early price and would have been a win only based those prices. 2/1 available at time of posting, which opens up the place market @ 4/6 with Skybet as a bet. This should be between Twister and Astromachia, both distance winners and last time out winners.

 

Assessed chance to place is 63.51% against place odds of 4/6 (60%) = Flat Kelly stake of 8.78pts

[skylark2009]

#10. 2.25 Doncaster - Dubai Beauty @ 2/1 Bet365 - 12.77pts

Some stats - 21 years (total runners =185)

Won last time out = 16 from 21 runnings (from a pool of 92)
First five in the betting = 21 from 21 runnings (from a pool of 110)
Returned SP <=4/1 = 12 from 21 runnings (from a pool 40)

I rate Dubai Beauty at 41.84% for this against available odds of 2/1 for a Flat Kelly stake of 12.77pts

 

#11. 3.00 Doncaster - Pilaster @ 11/8 Bet365 - 12.97pts

Pilaster is in excellent form this season with three wins from three, highest rated given the weights. Sizzling and Horseplay are the big dangers here for upsetting Pilaster.

Rated a 49.62% chance against odds of 11/8 for a Flat Kelly stake of 12.97pts

[skylark2009]

#12. 2.30 Lingfield - Layaleena @ 4/5 Paddy Power - 21.40pts

A Sir Michael Stoute trained Filly, who could very well follow up here (or at least place). Gives 6lb's to her rivals today for that win. The race she won has thrown up three next time out winners (plus three losers).

Rated a 65.07% chance against odds of 4/5 = Flat Kelly stake of 21.40pts.

[skylark2009]

3.40 Saint Cloud - Ummalnar @ 4/6  - 16.68pts

Assessed chance 66.67% against odds of 4/6 (Skybet) = Flat Kelly stake of 16.68pts

[skylark2009]

Results to date -

 

 

My actual bank for this is now down to 37.25pts (Full Kelly Bank = 48.17pts).

 

 

 

[skylark2009]

Two bets for today.

#14. 5.00 Newmarket - Hameem @ 1/1 - Kelly stake of 13.02pts (assessed chance of 56.51%)
#15. 5.05 Chelmsford - Swordbill @ 11/10 - Kelly stake of 14.34pts (assessed chance of 55.13%)

[skylark2009]

#16. 4.20 Leicester - Outbox @ 1/1  Paddy Power -Kelly Stake 12.5pts (assessed chance 56.25%)

A weak looking race which Outbox may be able to gain an advantage in. Could very well likely improve for his win last time out when running on strongly at the end. From a pace angle, Outbox and Canoodle will likely set the pace with many dropping behind, which Outbox (being a son of Frankel) my find advantageous.

 

Derived from my Excel DB list.
Outbox falls into a group of 48 qualifiers (since 1998) which produced 27 winners (56.25%), also 45 of these qualifiers also managed to place (1st, 2nd or 3rd).

[skylark2009]

Results to date -

 

Bet#14 Hameem returned BOG odds of 11/8.

 

 

LTO winners running on 25/09.

 

 

Some potentials by the look of it (price dependant).

 

 

 

[skylark2009]

#17. 3.10 Lingfield - Swordbill @ 1/1 - Flat Kelly Stake of 7.00pts (my assessed chance 53.50%)

Swordbill brought out quickly again and on for a potential 4th consecutive win here. Looks to be ahead of the handicapper with potentilly more to offer. Last three wins were on polytrack, so should be ok here. General Zoff looks a big danger here, beaten by Swordbill previously and is now rated higher (based on weight increase for Swordbill), General Zoff's jockey also looks a big assist here too.