- This topic has 34 replies, 9 voices, and was last updated 15 years, 4 months ago by
.
-
Posts
-
Erm, I’ve been thinking about this whilst I went shopping at Tesco’s.
I didn’t question Artemis’s figures, because it was Artemis.
Isn’t a 28% strike rate at just over 3/1 a Return On Investment of an astonishing 12%?
If so, to make £500 a week, one has to stake just over £4000 per week.
As in the uncensored version, nearly 100 selections are being made per week, one has to put about £40 to £50 on each selection.
If one has a bank of 100 units, that is a bank £5000, not half a million.
Yes?
I havent even thought about possible betting banks, something to ponder when the results are in
Gerald
You may be right. The figures were off the top of my head. I’ll have a more detailed look and report back.
Gerald
It is indeed 12%, which would be astonishing and would require a much smaller turnover and bank, as you say.
My error was (and is) to believe that such a return isn’t possible over 700+ bets. I, lazily and without checking, fell into my default mode of thinking that the best that could be expected would be between break-even(0%) and 5% profit on turnover, and based my figures on the average of this range(2.5%).
I thought £500 per week to be a fair target as a reward for the time and effort required to run the system. Most of us would settle for much less, just for the fun of doing it, so perhaps you wouldn’t need such a high turnover and bank. Not cowardice, but common sense because we couldn’t afford to lose more than £1000 without feeling the pain.
Just for info. I’m up to 634 assessed races so I’ll hopefully have some form of results by the end of the weekend.
The original interpretation of Kelly staking stated that you should wager the %age of your bank on each selection in relation to the size of your edge. This would mean that a 12% edge would mean staking 12% on every selection and re-adjusting the bank value and wagered stake with every bet.
This is mega risky and no-one is going to do that, unless you have a very high strike rate (70%+)
To overcome this fractional Kelly staking was dreamed up by someone. The original paper was re-interpreted to say you should try to win the edge of your bank at every bet and re-adjust your bank value after every bet.
So, if your edge is 12% and your bank is £1000 and the odds on your first horse is 3/1.
Your stake would be £1000 x 12% = £120 / 3 = £40
If this loses, bank (£1000 – £40) x 12% = ??? / odds of next selection.
Looks like a decent edge .. good luck ..
Oh dear, looks like we’re not only going to have an argument about whether Kelly is a good thing, but also what is Kelly.
I don’t recognise this betting the same percentage of the bank as your edge malarkey.
I’ll do a proper reply on Sunday, when I’m not working.
FYI
Kelly Criterion
.. there are a couple of links at the bottom of this article which you might find useful.
The original Kelly paper and a paper entitled Bayesian Kelly.
Afternoon all,
I assessed 672 in total over the past month applying the same system to every race. This has produced me with plenty of data so now time to assess the results.
From 672 races there were 181 winners (27%). Ave Price 3/1
From the 672 assessments 189 came out equal in the selection process with another horse. That meant the process had 483 races where there was 1 top pick. From these 483 races there were 106 winners (22%)At £5 stakes this would mean Stake + returns of £2196.20. losing stakes would be £1885. Producing a difference of £311.20
During the assessment the distances varied alot eg 5f 43y 6f 110y etc. These races were few and far between so I just concentrated on the results of those races where the distance was standard. (why is there a 5f 5f 43y race?
For 5f races there were 96 runs and 28 winners (29%) Ave Price = 5/2
@ £5 per race Stake + Rtns = £517.35. losing bets = £340. Difference £177.35For 6f races there were 83 runs and 22 winners (27%) Ave Price = 7/2
@ £5 per race Stake + Rtns = £494.35. losing bets = £305. Difference £189.35For 7f races there were 50 runs and 6 winners (12%) Ave Price = 6/4
@ £5 per race Stake + Rtns = £84.10. losing bets = £30. Difference £54.10For 1m races there were 68 runs and 13 winners (19%) Ave Price = 3/1
@ £5 per race Stake + Rtns = £281. losing bets = £275. Difference £6For 1m 2f races there were 71 runs and 21 winners (30%) Ave Price = 5/2
@ £5 per race Stake + Rtns = £377.90 osing bets = £250 diference £127.90For 1m 3f races there were 15 runs and 7 winners (46%) Ave Price = 4/1
@ £5 per race Stake + Rtns = £190. losing bets = £40. Difference £150.obviously I’ll need to complete a lot more races to have 700 for each distance. I’m affraid that this will not be possible.
If anyone who has conducted similar analysis I’d be interested to hear what ‘other’ areas I should look at.
Thanks
Afternoon all,
That meant the process had 483 races where there was 1 top pick. From these 483 races there were 106 winners (22%)
At £5 stakes this would mean Stake + returns of £2196.20. losing stakes would be £1885. Producing a difference of £311.20
Hi OFO,
My reading of the above is that at £5 stakes you end up losing £218.80.
Am I right?Kevin
I’d be wary that what has happened in the past will happen again in the future. Also, this is just for one month – May. It will vary during the year, I’m sure.
I’d also take on board Gingertipster’s excellent post in the Caledon thread. Once you start splitting things up, you’re bound to get some performing better than others, just through normal statistical variation.
Even so, my inclination would be to start betting. Low stakes though. If it works, you’ll generate a fair enough profit anyway, as there are so many selections.
Even so, my inclination would be to start betting. Low stakes though. If it works, you’ll generate a fair enough profit anyway, as there are so many selections.
My reading is that it is not profitable?
When considering clear top rated selections the S/R is 22% at average odds 3/1. This gives a loss of 12%.Kevin
Afternoon all,
That meant the process had 483 races where there was 1 top pick. From these 483 races there were 106 winners (22%)
At £5 stakes this would mean Stake + returns of £2196.20. losing stakes would be £1885. Producing a difference of £311.20
Hi OFO,
My reading of the above is that at £5 stakes you end up losing £218.80.
Am I right?Kevin
Hi Kev,
I thought the same as (1666.20-1885) = -£218.80. So what is the £311.20?
Hi Kev,
I thought the same as (1666.20-1885) = -£218.80. So what is the £311.20?
That is the total returned which incudes your stake ie 106 x £5
OFO you’ve included your original stake in with your P/L.
If your using decimal odds you need to deduct 1 before x your stake to find your P/L rather than Returns.
Regarding the analysis of results .. what do you want to check?
Also, it’s a bit odd to split the selections by distance, why would you want to do that?
- Total Posts 55
Hope you dont mind me throwing my hat in to the ring but……
If you looked at success based upon ‘distance run’ you will probably find that you are likely to see a great deal of fluctuation and would strongly suggest not going down that route. You are shaping your method around results to date but you need to make your spreadsheet dynamic and able to forecast trends.
You may find the most productive view regarding distance is where your selection has come from and where it is going.
Consider the age groups of the races you are looking at. Are you most successful with 2yo, 3yo, 3yo+ or 4yo+ races.
Your selection process may be better suited to a particular race type. Handicaps, Claimers, Conditions, Group Races
It may also be productive to monitor the race class and see if there is a noticeable % loss in higher or lower grades of race.
Most importantly there is a danger when you throw SP’s in to the equation of them distorting your thinking. I would look at your results from a winning % only. If you have a method which is giving you 25% you need to be backing at 4-1 for a profit. 30% 10/3. 40% 5/2. 50% 2/1 and so on. Backing smaller prices than your winning percentage will see you in the poor house!!
- You must be logged in to reply to this topic.