MC's NH System Test
Yesterday's selections lost.
Friday's selection lost.
|Lingfield||1.20||Tis Fantastic N||2.00|
I have now been running this for a month and on the loss so far it looks a little disappointing.
I will review the data over the next couple of days to see what it is telling me.
Monday's selection lost.
|Aintree||0.20||Quoi De Neuf||3.00|
|Chepstow||0.47||Elixir Du Gouet N||4.00|
|Chepstow||1.22||The Captains Inn 20||4.50|
Today's selections lost.
A winner and 2 losers yesterday for a profit of 2.8 points. Total loss 6.07 points.
Today's selection lost.
I have now reviewed November's data and the results are disappointing.
There have been 529 runners with 59 winners. The AE is 0.88.
Level stake profit at BSP is -44.99 points (-8%). Proportional profit is -14%.
The only encouraging aspect is if I only include those selections where it is the only selection in that race.
Of these there have been 101 runners with 15 winners. The AE is 1.07.
LSP is -4.02 points (-4%). Proportional profit is 9%.
The key rule of the system has held up well, DSLR > 27 has an AE of 1.02.
The 2nd key rule of the system has not fared as well, with horses aged less than 7 having an AE of 0.98.
This could be due to mean reversion which is always a risk with any investment system, be it stock market or horse racing.
It does raise a question which I don't think I have answered very well in the past. When do you have the confidence to start backing a system and when is the time to decide it is no longer working?
When do you have the confidence to start backing a system and when is the time to decide it is no longer working?
You 'pays your money and takes your choice' with this I think Michael.
The statistician's answer is, 'at the 99% confidence interval' for which there is a formula using the mean and standard deviation. A chi-square test is as good as any measure for horse racing results. However, I have seen systems show a chi-square of 100% for years and then bomb.
There are no guarantees but some people say, get to 95% chi-square then 'ride the wave' for as long as it stays there or thereabouts. I think you espoused a similar philosophy with level stakes betting. Bank the enhanced profits while the underlying A/E is greater than 1 then bale out when it starts to drop.
I think the only truly stable long-term methods are those where the advantage is 'baked in' mathematically as with most casino games and gaming machines. I guess the each-way odds in certain sized races when the favourite is short-priced would be a racing example.
I tend to agree it's difficult. As soon as I'm comfortable with a process I tend to start betting it. If it looks like it drops off I usually look for the decline to be over at least three months, but more likely continuous for six months to a year (unless it's very rapid but that's unusual). If I no longer have confidence I will still continue to monitor it for a period of time to make sure it wasn't just a downswing.
Yesterday's selections lost.
Yesterday's selection lost.
|Southwell||2.30||Full Of Surprises||6.50|
Sunday's selection lost.
Wednesday's selection won for a profit of 2.5 points. Total loss 8.57 points.
I will also use this trial to analyse the returns from the various betting options, i.e.
Best odds night before with BET365 or BOYLESPORTS (these 2 offer BOG the night before)
Best odds in the morning with BOG
A small data size but the results in November from the 49 winners I have had across a number of my systems are.
Best odds night before 141.54 points.
Best odds night before with BOG 163.69 points.
Best morning odds with BOG 161.39 points.
BSP 176.01 points.
So the conclusion so far is that BSP is the best option.
However if I analyse the data by odds the results are.
Odds < 9/2 - Best odds night before with BOG 85.69, BSP 78.26.
Odds >= 9/2 - Best odds night before with BOG 78.00, BSP 97.75.
So the revised conclusion is that BSP is the best option for odds of 9/2 and above, for odds below 9/2 the best option is to take the best odds the night before with BOG bookmakers.