Hello, i have been working on 2 systems, one which does use the last start winning criteria with some other changes, and one that work on long shots which have been working fine over the last 7 months

SYSTEM SELECTOR RULES SET FOR: LAST START WINNERS
Printed by Price Predictor at 22:54:20 on 20/02/2004

Race Distance: 1100 - 2400
Race Field Size: 8 - 14
Max No 1st Uppers: 8
Track Condition: GD
Horse Win %: 20 - 100
Horse Place %: 35 - 100
Career Starts: 8 - 999
Days to Last Start: 1 - 21
Runs this Preparation: 1 - 99
Last Start Finish Posn: 1 - 1
Max Distance Rise-Drop: 600 - 2200
Weight Change Min-Max: -20.0 - 10.0
Won or Okay at Distance: Y
Won at This Course: Y
Saturday Last Start: Y
Favourite Last Start: X
Predicted Posn Rank: 1 - 1
Predicted Odds: 0.0 - 15.0
SMBAWX 1234567

SYSTEM SELECTOR RULES SET FOR: LONGIES
Printed by Price Predictor at 22:55:08 on 20/02/2004

Day of Week: 7
Venue: SMBA
Track Condition: GD
Horse Win %: 10 - 100
Horse Place %: 20 - 100
Career Starts: 3 - 999
Runs this Preparation: 0 - 2
Last Start Finish Posn: 1 - 4
2nd Last Start Finish Posn: 1 - 10
3rd Last Start Finish Posn: 1 - 10
Max Distance Rise-Drop: 400 - 400
Won or Okay at Distance: Y
Predicted Posn Rank: 1 - 5
Predicted Odds: 0.0 - 20.0
SP Price: 10.0 - 999.9

If someone could check them for longer period than a couple of years it would be very appreciative. if tony could seeming you have the 5 years of data. thanks a heap, it may work out for all of us!!!

Adam, I ran your first system and got 77 winners from 348 selections at a loss of about 15 units, the winners including one at $33 and another at $19.

The second system I can’t run much further than you because of the lack of market data, but finished with 22 winners from 283 selections at a profit of 4 units. However the last (going backwards) bets were 42 outs followed by a 10.80 winner followed by 12 outs, and much longer runs of outs would have to be expected.

IMHO, common things happen commonly, and perhaps with enough common parameters to render some semblance of predictability.
However, long-shots happen chaotically and, although one hardly ever wins without a discernable reason, these reasons are normally too subtle and complex to be quantifiable according to the strictures of the likes of these programs.

I may be wrong but back-tested long-shot system hunting seems futile to me.

I may be wrong but back-tested long-shot system hunting seems futile to me.

Hi Tony,

I'd have to disagree with you there. I have a couple of longshot systems that appear to be going quite well.... They are just much more difficult to find and verify than usual.

Cheers,
Chris.

[/quote]

I sincerely wish you their continued success, Chris. However, I can't help musing that, come the inevitable mega run of outs, faith that the thing will come good merely because it happened to in the past may diminish in inverse and exponential relation to the mounting drawdown.

When you say they are more difficult to verify, does that mean you use means to do so other than the bottom line of a prolonged scan, statistically adjusted for the extreme results?

Cheers

Tony

Tony, are you able to shed any light on a system or 2 in which you use which have brang you success?

others that have worked fine over the last 7 months but again need to be long termed tested if someone gets some time

Race Field Size: 4 - 17
Max No 1st Uppers: 2
Track Condition: GD
Age of Horse: 1 - 7
Career Starts: 3 - 999
Barrier Position: 1 - 12
TAB Number: 1 - 12
Runs this Preparation: 1 - 99
Last Start Finish Posn: 1 - 3
2nd Last Start Finish Posn: 1 - 3
3rd Last Start Finish Posn: 1 - 3
4th Last Start Finish Posn: 1 - 3
Won or Okay at Distance: Y
Won at This Course: Y
Predicted Odds: 0.0 - 3.0

thanks

Shades of the old Dark Horse for which I parted with 50 bucks umpteen years ago.
Not many bets, Adam, but a pretty good result.
High strike rate lends itself to a staking plan?? Are you still using Power of 10, Darky?

Meetings considered : 2079
Win Strike Rate/Seln. : 42.4%
Plc Strike Rate/Seln. : 72.4%
Average Win Dividend : $2.57


Races Bet : 240
Races Won : 103
S.R./Race : 42.9%
Outlay ($): 243.00
RETURN : 264.70
$ Profit : 21.70
% P.O.T. : 8.9%

When you say they are more difficult to verify, does that mean you use means to do so other than the bottom line of a prolonged scan, statistically adjusted for the extreme results?

Hi Tony,

the summary of results that Price Predictor gives is only the start of what I do to try to verify a system. I look at a number of different aspects of the results in general, and some very specific ideas as well.

I graph the cashflow, look at month-to-month performance (s/r and pot), calculate longest run of outs and maximum drawdown, recalculate the s/r and pot by missing a few big winners and so on. When looking for longshots, it's not as applicable to drop the top few dividends and recalculate, since you are deliberately trying to target longer priced selections. One thing that I do look at is the dividend for every winner.... I am a lot happier if there is a steady stream of $10 to $30 winners rather than the occasional $40 or $80 winner. I also like to see an occasional $4 or $5 winner as well.

I've found many systems that have very nice looking S/R and POT's at the end of the run, only to find that all the winners came in a short period of time, or all the profit came from a short period of time.

The other part of verifying a system comes from "forward testing"... following the selections on paper for a few months. This is difficult with longshot systems as you need to wait a lot longer for the system to look promising.... and you might invariably miss a really big winner.

When betting a system, I expect the longest run of outs and maximum drawdowns to double at some stage during my betting (and they generally occur early in the sequence), so I won't start to get concerned until this happens. Structuring the bank to withstand such a drawdown has saved me many times, and one of my better longshot systems relied on almost every cent of it's own bank before it bounced back.... that bank has tripled in the last month :smile:

Cheers,
Chris.

Thanks Chris,

I used to generate an equity curve from the excel file ppp system tester provides.
However I lost the relevant formula a couple of computer crashes ago and am excel illiterate. I think I originally plagiarised it from a commercial service boasting past results.
Would appreciate help with it as it is a bit of a must.


"When betting a system, I expect the longest run of outs and maximum drawdowns to double at some stage during my betting ..."
You are meaning double compared with back test period? What raw pot on back test do you cosider minimum to warrant further investigation of a potential system?

Appreciate your info, hope the long shots keep doing the right thing by you !


Tony

Hi Tony.Yes I do .From BET SMARTER AND WIN.
Cheers.
darky