I want to propose a system which I have now backtested from early January 2006 until beginning of June which includes both All Weather and Turf meetings.
RULES
1. Race Distance must be under 2 miles or 16f
2. Horse must have highest Master Rating (can be shared top);
3. Horse must have had a minimum of 3 Runs;
4. Horse must have the highest rating from the last 3 runs (can be shared);
5. Horse with the Highest Master and Highest Rating from last 3 runs gets selected provided they are equal. [PLEASE NOTE THIS AS VERY IMPORTANT] If there is another qualifier then the one with the highest speed LTO is chosen. If there is still a tie then the one with the highest M+A is chosen. If there is still a tie then no selection (highly unlikely);
All Weather
6A Selections only valid in classes 3-6 inclusive
7A Selection can only be taken if SP price is between 15/8 to 7/1 inclusive
Turf
6B Selections only valid in classes 3-5 inclusive
7B Selection can only be taken if SP price is between 5/1 to 10/1 inclusive
Because of SP stipulation you must wait until 60 seconds before the off to take the bet.
Taking prices early is an incorrect strategy even if a higher price is obtained as all analysis has been made against SP pricing.
By using Betfair the SP price should be easily beaten by between 10-20%. I know it can be difficult when prices are hovering at the extremes of the lower/upper ranges but you must wait until close to the off and be patient and disciplined.
LOGIC
By using the highest speed from the last 3 runs and equating it the highest Master rating in the field we are ensuring that one of those runs made the Master rating in the last 12 months.
Next we may be forgiving a poor run LTO and are hoping that the horse can repeat its performance again by beating the rest of the field with the fastest speed of all runners form their last 3 runs.
By using the various filters by surface, Distance, Ran LTO, Class and SP ranges we are trying to eliminate all those selections that do not add to the profit or if they do it is a poor rate that it is better to leave alone.
KNOWN ISSUES
There are some missing races in May that I need to add to my database to assess. Also at the beginning of January and some dates in May I have missing LTO day’s data which I need to fill in. As soon as I've done that I will post up any changes. Hopefully it will not cause too a dramatic change.
The number of qualifiers is only over 100 in 5 months; this is not statistically safe as it introduces a margin of error until the population size is at least 200 and spans at least 3 seasons. Unfortunately this data and period is not available for backtesting.
As a result of having little backtested data any money wagered should be very small and something you can afford to lose until more data is added or can be backtested.
PERFORMANCE (EARLY JAN TO JUNE 01, 2006)
DRAWDOWN = -3.00
LONGEST LOSING RUN = 7
PEAK TO TROUGH = -10.50
SURFACE; WIN/SELECT, SR%, £1 P/L, ROI%, AO, A/E
ALL WEATHER; 28/81, 34%, +49.46, 61%, 4.65, 1.61
TURF; 9/24, 37%, +45.00, 187%, 7.67, 2.87
TOTAL; 37/105, 35%, +94.46, 93%, 5.39, 1.90
MONTH; WIN/SELECT, SR%, £1 P/L, ROI%
JAN (PARTIAL); 1/7, 14%, -4.25, -61%
FEB; 10/23, 43%, +21.21, 92%
MAR; 8/29, 27%, +11.25, 39%
APR; 14/26, 54%, +40.75, 157%
MAY (INCOMPLETE); 3/18, 17%, +18.75, 105%
JUN; 1/2, 50%, +7.00, 350%
SR% = STRIKE RATE %AGE WIN/TOTAL
£1 P/L = PROFIT TO LEVEL STAKES USING RETURNED SP
ROI% = RATE OF RETURN %AGE PROFIT/TOTAL
AO = AVERAGE ODDS IN DECIMALS
A/E = ACTURAL WINS DIVIDED BY EXPECTED WINB (MORE THAN 1 IS DESIRED)
If anyone has any comments or could see what improvements could be made then I would appreciate hearing from you.
VC......Well done again...usual impressive stuff.ROI of 95% should keep everyone happy.I am off with a rope to find a tall tree after my system here produced 7 losers yesterday.
Hi VC, wow what a lot of info! You have been busy but it looks to have been worthwhile. Well done.
A couple of questions for you if you don't mind.
Do you back in all races including Maidens, 2yo races?
Can you explain 'drawdown' please
And peak to trough.
I hope I'm right in saying that this system has made 94.46 pts since January?
Excellent.
Keith
Dylan
Your system can be improved with some filters and I have suggested some in your thread and it can complement this system very well as very few qualifiers will be picked simultaneously.
Keith
The class filter elimiantes all the deadwood, so in answer to your question if there is a Maiden in the class range for each surface then it will be included.
Drawdown should have been stated as Initial Drawdown which means how much money (in terms of points) you would have had to lay out from the start of the system before it became profitable and you would not have had to use any of your money. So in this case it was 3 points. so multiply that by wahtever staking you use.
Peak to Trough is a measure of the worst case drawdown from a peak profit to a trough profit if you did not take out any profits. So in this case there were a string of losers intersperesed with winners but the overall result was that you would have been down over 10 points. this is where you would utilise this number to design a betting bank. I would recommend at least double or possibly even treble this figure.
The total points are level stake at returned SP. With BF prices close to the off they would be higher.
I have not traded this system until very recently as I've had to accumulate enough data to perform backtesting. I am not satisified with the number of qualifiers as being statistically safe so only very small stakes should be wagered. I would like to have analysed all races for at least 3 seasons and have at least 200 qualifers, the more the better to reduce Margin of Error.
Hi VC
Very good research and stats, well done.
Do you have the list of qualifiers so that I can look at them from different angles, ie trainer in form at the time of selection, drop in class, well drawn etc..
Many thanks
Monster
Hi Monster
I want to finalise all the missing data as given in 'Known Issues' so that I can firm up the rules hopefully with more qualifiers for the period that I have full data. I will then detail all the qualifiers so you can research further. I hope I can do this over the weekend (racing permitting!).
Hi VC
Cheers, I will also be able to check the Betting Forecasts that were available the night before racing took place.
Good Luck
Monster
I forgot to mention in the rules that for both surfaces the LTO rating can be no longer than 42 days.
Here are the qualifiers and results for June to W/E 04 June 2006:-
01/06/06, 3 Robustian(7), 8:55SANDOWN(TR), 15/2, 2ND, -1.00
01/06/06, 10 Taboor(5), 4:30YARMOUTH(TR), 8/1, 1ST, +8.00
02/06/06, 8 Starboard Light(2), 4:25WOLVERHAMPTON(AW), 5/1, UP, -1.00
03/06/06, 6 The Leather Wedge(7), 2:20MUSSELBURGH(TR), 10/1, UP, -1.00
03/06/06, 8 Speedy Sam(12), 6:20SOUTHWELL(AW), 5/2, 1ST, +2.50
03/06/06, 2 Smile For Us(1), 7:50SOUTHWELL(AW), 6/1 UP, -1.00
SUMMARY
W/T; 2/6, 33% SR, PL£1; +6.50, 108% ROI
I have found a couple of new problems, the first that will affect the whole of May (which is in addition to the missing data I posted up before) concerns all AW cards where none were taken into consideration (faulty coding on my part).
The other problem which will cause some changes for sure, is that when there was a N/R the filters should have been applied to the remainder of the field who were still running. This was not done at all, and I have to regurgitate all the qualifiers/results from January onwards.
At least I know June is completely correct as posted above.
Hi VC
Should Sir Percy be counted as he was Top master rated, Top LTO and was in the correct Sp Band 6/1
Good Luck
Monster
Monster
The Derby was a class 1 event thus was not applicable to the system.
Hi VC
Noticed another, Solent 5.05 Epsom 5/1
Good Luck
Monster
Hi Stuart
Posts crossed Solent's race was a class 2
Cheers
Monster
Hi Monster
Turf
6B Selections only valid in classes 3-5 inclusive
The stats for this look really good as a system but warnings re the smallish sample set.
Unfortunately I cannot play as I have no access at work so cannot bet near the off.
Good Luck
Stu
The Non Runner issue has had a significant impact thus far and it is not positive. It has knocked well over 30 points from the results though still in profit by almost 60 points for 2006.
There were so many races affected and obviously the 1st choice qualifiers were in the Non Runner list, the next best that were chosen obviously did not perform as well as those chosen where there were no Non Runners - maybe that should be a filter?? No that will not be good as it's best to select the top in the field that are actually running. However it can get very complex as the N/Rs can be at any time even getting withdrawn at the last moment before race start.
In light of this change, I have to re-evaluate all the filter parameters after finalising all missing data to be inputted from Jan to May inclusive. I will post up my findings as soon as I have them. So I would advise anyone contepleting following the rules to hold off for now even though June has prooved very profitable. I have certainly had the benefit but it seems to be down to a good luck more than anything else at this stage.
Hi VC
Been following this thread with interest. Not sure whether I'm being thick..... You make the selection following the rules. If your selection doesn't run then its a no bet.
If making a second selection after the first becomes a non-runner causes a big drop in profits, then write in the rules no second selections.
(Probably I'm missing the important point somewhere....)
Cheers
You are not being thick at all, it was a musing of mine in my previous post. However I have not completed all the changes that I need to do and will then re-evaluate the whole period again to see what's best in terms of filtering.
I have the ability to filter even further than I have ie by Number of Runners, Going, Race Age, Stakes/Handicaps, Rider Type, Race Gender, Race Type (Maiden/Novice/Conditions/Classified/Banded/Listed/Group/Claiming/Selling/Open). However I do not want to apply these yet as it would decrease even further the number of potential qualifiers which so far is low anyway.
I'm always looking for quality over quantity that means very low losing runs and peak to trough drawdowns - I can't accept more than 10 as a 1 in a lifetime event and a very high SR% and ROI% with no back to back losing months. If I had more data to work on (usually I like 10 years worth but minimum 3) then I would want at least 200 qualifiers to have a minimum Margin of Error in my results.
I hope I can make all the relevant corrections and crank out a decent set of results so that I can post them up. It should be done hopefully in the next few days or so.
I have had a number of good priced winners using the ratings and I think this HSL3 system's results are very impressive. However I am confused with regard to the rules 6a, 7a, 6b and 7b selections. I would be grateful if someone would please explain what these mean an/or where do I find the information
Hi John
6a / 6b races are classified 1 through to 6, with Class 1 races being for the elite horses and Class 6 for dregs. The class of a particular race is invaiably included in the info at the top of the race card. Ian puts it in the header on the Inform ratings - in brackets. Eg. Todays 2:20 at Carlisle is a Class 6. 2-20 CARLISLE 5f ,10 run
NORTHERN RACING DISCOUNT CARD MEDIAN AUCTION MAIDEN STAKES (6) 2YO
I think your other question was about SP. SP is starting price - which is the final bookmakers odds available at the off. Its easy to find out the SP after the race as these are the odds quoted in all results. Impossible to precisely predict the SP before the race, but as VC says, if you take the price 60 seconds before the off you will probably be there. Alternatively you can take the SP price when you place the bet - but that means you don't actually know what odds you have got and whether they fit the HSL3 rules until after you have done your dough.
Cheers
The rules concerning Class is BHB's rating of the race quality. The number is found in IR's race cards in brackets, ie (6) in the following race from yesterday:-
2-20 BRIGHTON 6f ,5 run
BLUE CHIP ORIGINAL EBF NOVICE MEDIAN AUCTION STAKES (6) 2YO
the rules concerning price is the returned SP (Starting Price) for every runner after the race completes. so from yesterday's results for the first race in the Racing Post:-
1 Bahama Mama (IRE) (4) J Noseda 2 9-1 Shane Kelly 1/6F — —
2 3½ Pires (5) M R Channon 2 8-12 Sam Hitchcott 6/1 — —
3 6 Daring You (1) P F I Cole 2 8-12 Jose de Souza 18/1 — —
3 ran TIME 1m 8.90s (slow by 0.20s) TOTAL SP 105%
NON RUNNERS: Going Straight (IRE) ,Proper (IRE)
The retruned SP's are 1/6, 6/1 and 18/1 respectively. The rules require that a bet is only taken close to the off within the stipulated price bands at the prices provided from the Ring as displayed by SIS and bookies. These will never be 100% accurate compared to the returned SP but it is the best information that is available.
