Find Selections Usi...

Clear all

# Find Selections Using Impact Values

Page 2 / 3
(@mikeywilding)
Joined: 10 years ago
Posts: 1395
25/02/2020 12:57 pm

SHorAvG

Rank 1: A/E 0.95 (14% SR)

Rank 2: A/E 1.08 (14% SR)

Rank 3: A/E 1.04 (12% SR)

Rank 4: A/E 0.92 (10% SR)

5278 CFR

Rank 1: A/E 1.13 (24% SR)

Rank 2: A/E 0.95 (16% SR)

Rank 3: A/E 1.07 (15% SR)

Rank 4: A/E 0.99 (11% SR)

PFP

Rank 1: A/E 1.04 (18% SR)

Rank 2: A/E 1.10 (17% SR)

Rank 3: A/E 1.08 (14% SR)

Rank 4: A/E 0.94 (11% SR)

SP RANK (CONTROL SET)

Rank 1: A/E 1.05 (33% SR)

Rank 2: A/E 1.02 (19% SR)

Rank 3: A/E 0.98 (13% SR)

Rank 4: A/E 1.03 (10% SR)

If we start simply, and just combine the two most obvious factors:

1) PFP Rank 1st, 2nd or 3rd

2) 5278 CFR 1st rank

The A/E is 1.23 with a strike rate of 30%. In 2019 it produced 205 winners from 684 selections and made a profit of +98 points to flat stakes after a 5% commission.

To proportional stakes a profit of +25 units would have been made.

I've attached the charts in the following posts.

Now of course, there's an element of fitting to data here. But it's a strong start. Would you like me to continue?

(@mikeywilding)
Joined: 10 years ago
Posts: 1395
25/02/2020 12:58 pm

The chart for the proportional stakes attached.

(@andrewp)
Reputable Member
Joined: 1 year ago
Posts: 398
25/02/2020 6:57 pm

So far, so good Michael, please continue.

Classic level stakes (overstaking) and proportional stakes (sensible staking) charts :0)

Just for comparison, am I right in thinking,

1. Level stakes POT = 98/684 = 14%

2. Proportional stakes = stake to RETURN 1 point (rather than to make a target profit)

3. Proportional stakes POT = 23% before commission (A-E)

(@mikeywilding)
Joined: 10 years ago
Posts: 1395
26/02/2020 11:08 am
Posted by: @andrewp

So far, so good Michael, please continue.

Classic level stakes (overstaking) and proportional stakes (sensible staking) charts :0)

Just for comparison, am I right in thinking,

1. Level stakes POT = 98/684 = 14%

2. Proportional stakes = stake to RETURN 1 point (rather than to make a target profit)

3. Proportional stakes POT = 23% before commission (A-E)

1. Level stakes POT = 98/684 = 14%

Yes, that's right.

2. Proportional stakes = stake to RETURN 1 point (rather than to make a target profit)

Yes that's right.

3. Proportional stakes POT = 23% before commission (A-E)

I make it 15.26% after a 5% commission.

(@mikeywilding)
Joined: 10 years ago
Posts: 1395
26/02/2020 12:00 pm

Let's look at a breakdown of the odds using proportional stakes:

 Odds Count of Win Count of Lose Sum of Profit Strike Rate 1.00-1.499 9 15 -0.41 38% 1.50-1.99 17 34 -0.54 33% 2.00-2.99 33 74 3.16 31% 3.00-3.99 26 62 3.42 30% 4.00-4.99 28 55 6.32 34% 5.00-5.99 15 46 0.51 25% 6.00-6.99 8 32 -1.86 20% 7.00-7.99 12 31 0.87 28% 8.00-8.99 10 22 2.14 31% 9.00-9.99 8 16 1.74 33% 10.00-14.99 29 40 10.38 42% 14.99-29.99 7 31 -0.67 18% >=30.00 3 19 -1.01 14%

We can see here that there is a sweet spot in the odds between 2.00 and 4.99, which has produced +12.90 units profit over 278 bets, with 87 wins for a strike rate of 31.29%.

That's just under one bet per day.

There's also the possibility of a lay approach for odds-on horses at 1.50 or less, but would need more work or manual analysis to make it profitable. As well as a higher odds approach for horses with odds. between 8.00 and 14.99. These horses have produced a profit of +14.26 over 125 bets, with 47 winners and a 37.60% strike rate.

We don't really have enough data to be confident in the higher odds approach. Ideally I like to have a minimum of 100 winners before making a decision. Either we could track this live, or get a few more years historic data to see if it's held up over a longer period of time.

When to start betting a strategy is a difficult decision for most bettors. To get enough data statistically requires a lot of selections and usually a long time. For most people this isn't practical unless they're using a very high turnover mode. However, we want enough data so that we're not purely guessing, which is why my preference is for 100 winning bets.

However, we can also use an approach to our betting where we start with a betting bank we're happy to lose and re-invest all profits in the aim to double it as fast as possible. Once it's been doubled then take out the original bankroll and reduce the risk and speed of growth, as well as removing a percentage of monthly profits.

So far we've only used two elements, from two factors, of the three factors we chose at the beginning. We could look at at building out the lay approach, getting more data for the higher odds, or looking at other elements of the factors we chose?

(@michael-clarke)
Prominent Member
Joined: 5 years ago
Posts: 784
27/02/2020 12:51 am

Michael, a very good thread, it will be interesting to see the results for 2020.

I am a little dubious about the conclusions about the odds, it does not seem logical to lay a selection at odds of less than 1.50 and back them at odds of 2.00 or above.

Also I am not quite sure how you make a profit with the horses in the 7-7.99 range with only 2 winners form 42 selections. The 40 losing selections would require proportional stakes of at least 5 points with only 2 points recovered by the 2 winners, unless I am misinterpreting the calculations.

(@mikeywilding)
Joined: 10 years ago
Posts: 1395
27/02/2020 9:56 am

@michael-clarke thank you Michael. I just went back and checked the pivot table because, as you said, you can't make that profit from 2 winners out of 42 runners. Turned out there was an issue with the count on the winners. The table above has now been updated and edited. The figures for the 7.00 to 7.99 range were a loss of -5.35 and a profit on the winners of 6.30.

The results so far in 2020 are:

Expected Winners: 30
Actual Winners: 38
A/E: 1.26
Selections: 124
SR: 31%
Proportional Profit: 6.02
Flat Profit: 11.71

As I always say, I would use these as the shortlist and manually cross-check them before betting them.

(@andrewp)
Reputable Member
Joined: 1 year ago
Posts: 398
27/02/2020 9:01 pm

Michael,

The breakdown by odds range shows 313 winners, 659 losers (so 972 runners). This is different from the initial summary of 205 winners, 479 losers, 684 runners.

(@michael-clarke)
Prominent Member
Joined: 5 years ago
Posts: 784
28/02/2020 12:12 am

Michael, is it possible to post the selections the night before racing so that we can see how they perform "live".

(@mikeywilding)
Joined: 10 years ago
Posts: 1395
28/02/2020 12:32 pm
Posted by: @andrewp

Michael,

The breakdown by odds range shows 313 winners, 659 losers (so 972 runners). This is different from the initial summary of 205 winners, 479 losers, 684 runners.

Yup, just looking now to see what I've done in the pivot table. I knew I preferred databases for a reason 😀

(@mikeywilding)
Joined: 10 years ago
Posts: 1395
28/02/2020 12:39 pm

I'm just going to redo these sums without a pivot table so I know I've got them right.

(@mikeywilding)
Joined: 10 years ago
Posts: 1395
28/02/2020 1:02 pm

Right, I have now done it manually, no pivot tables. We are two horses short because we don't have odds for them, but they are losing bets.

The figures are pretty much the same, the strike rates have marginally increased and profit slightly decreased. Thank you @andrewp for spotting.

(@mikeywilding)
Joined: 10 years ago
Posts: 1395
28/02/2020 1:03 pm
Posted by: @michael-clarke

Michael, is it possible to post the selections the night before racing so that we can see how they perform "live".

I can't guarantee to be able to post them every day, but maybe a member can offer do track them live. I will be doing a Facebook Live event about this at around 2pm, keep an eye on emails if you're free to join.

(@andrewp)
Reputable Member
Joined: 1 year ago
Posts: 398
28/02/2020 2:01 pm

We are two horses short because we don't have odds for them, but they are losing bets.

That's a shame as it affects the 'expected winners'. I may have the Betfair prices if you let me know the runners.

The figures are pretty much the same

I presume this means the summary figures and the breakdown by odds band have the same number of winners, losers, runners? When you get a moment could you post those up or upload the raw csv perhaps?

(@mikeywilding)