It's something I have never considered before but it's started to float around in my mind. Is there any research into how quickly prices fall (or drift) all other things being equal by which I mean ignoring the immediate changes when a horse falls or refuses.
I'm expecting it to be a bit like an upside-down exponential curve similar to the deterioration of the Draw in Soccer Mystic when no goal is scored. Does anyone know of any values/formulae/theories available. Yes, every race is different but so is every football match but that doesn't stop theories of how an average game might play out.
I'm looking at this because of the number of in-running bets not being matched so I'm thinking is there a way to estimate either a slightly earlier trigger to make sure your bet at the target price is matched or trigger as normal but set the target lower? Easy to do but am looking to judge what would be a workable margin between Trigger and Target.
The rate of Horse Racing price change in-running
I doubt whether comparing a football match and a horse race is very helpful. In a football match there are three outcomes win/draw/lose in a horse race there are two outcomes for the horse win/lose or numerous outcomes, if you look at the race as a whole and consider the other horses.
With the declining draw, the draw is in pole position and the chances of it happening obviously increase as the final whistle gets closer. With a horse race, being in pole position is not necessarily an advantage, as the front runner is often caught by other horses who have run more tactically. The odds are affected by how well the horse is running, so I can't see how any rigid formula could be applied.
With the declining draw, the draw is in pole position and the chances of it happening obviously increase as the final whistle gets closer. With a horse race, being in pole position is not necessarily an advantage, as the front runner is often caught by other horses who have run more tactically. The odds are affected by how well the horse is running, so I can't see how any rigid formula could be applied.
-
JuiceyJones
- Posts: 168
- Joined: Wed Sep 02, 2020 3:00 pm
I saw a dashboard on Youtube last year.
It basically had recorded prices for all runners over 5 second intervals for all races over the last 15 years (i think). The dashboard would monitor all of the days races on betfair then at about 1-2 minutes before the off look through its database to find close matches to a race was about to start. Historical races that were close in price for each horse, trip, going, class etc. Then give average price moves from its historical data.
I have no idea if it worked or would work. Pretty sure you can do most of it in BA.
It basically had recorded prices for all runners over 5 second intervals for all races over the last 15 years (i think). The dashboard would monitor all of the days races on betfair then at about 1-2 minutes before the off look through its database to find close matches to a race was about to start. Historical races that were close in price for each horse, trip, going, class etc. Then give average price moves from its historical data.
I have no idea if it worked or would work. Pretty sure you can do most of it in BA.
-
ShaunWhite
- Posts: 10530
- Joined: Sat Sep 03, 2016 3:42 am
Rate of change and volume weighted momentum (where the rate of change has 'mass' , aka volume) are both valid signals to put in your algo. Usual problems choosing an appropriate time span for the delta though.firlandsfarm wrote: ↑Sun Aug 17, 2025 5:12 amIs there any research into how quickly prices fall (or drift)
Aggressor imbalance is a good in-running indicator too, but I'm not sure if BA can identify tv changes per price, and then look back at whether that was a back or lay price at the time it changed.
-
firlandsfarm
- Posts: 3337
- Joined: Sat May 03, 2014 8:20 am
Thanks for all your contributions guys ... I never thought it would be straightforward nor easy. As I said I'm looking for a way to try and improve bet matching in-running.