Substantial Model Improvements
22/05/06 - Punter's Genie
With the recent poor weather battering some of our predictions we have spent quite
a bit of effort over the last week on the question of updating the predictions during
the day. There are advantages and disadvantages in doing this - the main disadvantage
being that whichever prediction we report on in our results, you can be sure that
some people will be upset because they didn't get a chance to see it and therefore
will claim that we are not reporting on our results accurately. This is a valid
complaint, but only if you don't tell people what is going on! We have seen many
sites where many successes are due to tweaks in the betting advice near the start
of the race. Rest assured, if we do change anything about the way we report on our
results, we will make every effort to make sure that this is clearly indicated on
the site. We pride ourselves on being open and honest in a sometimes less than honest
game!
Now to the advantages. The fact that up-to-date information is going to be better
than old information hardly needs clarifying. We produce the predictions the day
before the races using whatever information is available to us, including race cards,
forecast prices and so on.
One of the main benefits of producing the predictions
the day before the races is that it allows our members to get to the market early
- usually before most of the other punters - and get the best prices. The downside
of this is that overnight changes are not taken into account and can, sometimes,
affect the outcome of a race. These changes can include the going, the number of
horses in a race, the jockeys on each horse and so on. However, having said that, and all things being equal,
we have a pretty good track record on the flat, and last year's efforts will back
that up.
We know we can do better though - much better. We already know that we perform much
better when the overnight conditions do not change, the jockeys do not change and
there are very few non-runners. If we could apply this same principle of least possible
change to all races, we should be able to make more profit per race than we do now.
And that is exactly what has happened in testing last week. For example, on one
day last week, by updating our data and running the predictions again just after
midday, we made
11 points more on paper using the Win model only than we would have
made with the standard predictions (see
below
for the comparison). In fact, every day we tested we made multiple-point
improvements on the models' performance. This is particularly important for the
Win models where we are aiming at a profit of, say, around 5 to 10 points a day.
Just 2 extra points would be an increase 20-40%. Hardly to be sneezed at, and we
are confident we can improve on that.
So, what are we going to do about it?
Well, we are going to pass these improvements
on to you, our members. Our plan is to start updating the predictions twice daily
- once at about midday for the afternoon races, and once at about 4.30pm for the
evening races. Once this processes have bedded in and we've done some more research
we hope to make these updates "race-based" so that we can update each race individually,
multiple times during the day, and most importantly soon before the off so that
we can give the best possible predictions for the race using all of the information
that is available to us.
On the reporting side, for our results we will use the latest predictions available
on the site before the race. Although we understand that not everyone can keep
an eye on the site to get the updated predictions, we would hope that most people would
be able to check our predictions just before you put your bets on. Some people,
of course, wait until just before the race before placing their bets and they will
find most benefit in these changes.
We think that this change is for the general good of Punter's Genie and, most importantly,
you. The "evening-before" predictions will still be as valid as they are now, but
we hope that the updated predictions made during the day will substantially improve
your profits!
When we make these changes (possibly this week) we will make sure that it is clearly
indicated on the front page of the site.
If you have any questions about this or any other aspect of Punter's Genie, please
do not hesitate to get in touch by e-mailing us at
info@puntersgenie.com.
Comparison of current versus proposed models
for Saturday 20th May
The table below shows you just how well the new method worked for one particular
day during testing. Anything that has changed from the overnight predictions is
highlighted in light green in the "Horse" and "Stake" columns. The profits for each
race are shown in the "Profit" with the totals for each system at the bottom. As
you can see the new system outperformed the current one by more than 11 points!
| Race Information |
|
Results fron the Current System |
|
Results from System with mid-day updates |
| Course |
Time |
|
Horse |
Stake |
Place & odds |
Profit |
|
Horse |
Stake |
Place & Odds |
Profit |
|
Newbury |
13:35 |
|
Conquest |
2 |
5th (Evens) |
-2 |
|
Conquest |
2 |
5th (Evens) |
-2 |
|
Nottingham |
13:55 |
|
Cashel Mead |
0 |
1st (7/2) |
0 |
|
Cashel Mead |
1 |
1st (7/2) |
3.5 |
|
Newbury |
14:10 |
|
Distinction |
1 |
1st (3/1) |
3 |
|
Distinction |
2 |
1st (3/1) |
6 |
|
Nottingham |
14:25 |
|
Sphinx |
1 |
1st (4/6) |
0.67 |
|
Sphinx |
1 |
1st (4/6) |
0.67 |
|
Thirsk |
14:35 |
|
Eliza May |
0 |
1st (11/4) |
0 |
|
Eliza May |
1 |
1st (11/4) |
2.75 |
|
Thirsk |
15:10 |
|
Virgin Islands |
2 |
3rd (5/4) |
-2 |
|
Virgin Islands |
2 |
3rd (5/4) |
-2 |
|
Newbury |
15:20 |
|
Pearly King |
1 |
1st (9/2) |
4.5 |
|
Pearly King |
0 |
1st (9/2) |
0 |
|
Nottingham |
15:35 |
|
Indian Maiden |
1 |
2nd (13/8) |
-1 |
|
Cape Columbine |
0 |
3rd (85/40) |
0 |
|
Thirsk |
15:45 |
|
Heaven Knows |
1 |
1st (7/4) |
1.75 |
|
Heaven Knows |
1 |
1st (7/4) |
1.75 |
|
Newbury |
15:55 |
|
Chocolate Caramel |
0 |
5th (9/4) |
0 |
|
Cruise Director |
1 |
4th (3/1) |
-1 |
|
Nottingham |
16:10 |
|
Prince Of The May |
0.5 |
10th (8/1) |
-0.5 |
|
Trouble Mountain |
1 |
1st (11/2) |
5.5 |
|
Thirsk |
16:15 |
|
Bluebok |
1 |
11th (11/2) |
-1 |
|
Bluebok |
1 |
11th (11/2) |
-1 |
|
Newbury |
16:25 |
|
Red Evie |
1 |
1st (11/2) |
5.5 |
|
Spirit Of Arosa |
1 |
2nd (7/1) |
-1 |
|
Nottingham |
16:40 |
|
Follow The Colours |
0.5 |
NR |
0 |
|
Diamond De Triano |
1 |
1st (5/1) |
5 |
|
Newbury |
16:55 |
|
Pirouetting |
2 |
4th (3/1) |
-2 |
|
Pirouetting |
0 |
4th (3/1) |
0 |
|
Newbury |
17:25 |
|
Mashaair |
2 |
3rd (5/4) |
-2 |
|
Mashaair |
2 |
3rd (5/4) |
-2 |
|
Thirsk |
17:50 |
|
Stolen Glance |
1 |
2nd (2/1) |
-1 |
|
Stolen Glance |
1 |
2nd (2/1) |
-1 |
|
|
|
|
|
|
|
|
|
|
|
Total Profit |
+3.92 |
|
Total Profit |
+15.17 |