New Models

[nick257]

This week we’ve been working on a new prediction model approach, which has gone live today, Saturday 26th September. You should not be betting just yet… (apologies for the false start!).

The system will now choose, automatically, between 2 different prediction model types each day.

• The first, is our original approach, using over 100+ predictors (independent variables), including the public odds.
• The second model type, uses 3 fewer predictors, which are all related to the public odds ([Starting_price], [Forecast_Price], and [position_in_betting_order]).

The thinking behind the new approach, is that whilst previously we included the Public odds as part of the prediction models, currently, these predictors are adversely affecting the models – so we’re going to try and predict winners without using the public odds at all.

You will notice a few changes in the way the system picks runners…

It will no longer pick 1st, or 2nd place in betting order runners, all the time. You will see it occasionally pick runners further down the betting order.

The Stop-Loss…. The average Win Rate for the new models is lower, around 25% (previously it was around 31%). The average price will be higher though… This means, when we’re using the Models without Public Odds, we’ve had to increase the stop-loss to 60 points per day, and 18 points per course (previously it was 30 points per day, and 16 points per course). The stop-loss values will be changed over the next few days/weeks, as we get more results in… and are able to fine tune it.

When the models include the public odds, the Stop loss will be 30 points per day, 16 points per course.

For the time being, we will not be recommending betting with the new models – until they’ve had a chance to prove themselves. We have been testing them for the last 7 days in our development environment – and they did make a profit.

I’ll be changing the website later – so you can see which types of models we’re using each day…

That’s all folks,

Cheers
Nick

[Author]

Posts