Scalability

[firlandsfarm]

Scalability is often mentioned when it comes to strategy design but how do you anticipate it and determine it's boundary. Is something that works with £1 stakes going to work at £10 and if it does will it work at £100?

We spend a lot of time and effort researching strategies so it would help if we could estimate the scalability of a strategy before embarking on the hours. I was wondering if people have any ways they examine the extent a trial strategy may be scalable before spending the hours refining it?

[sionascaig]

For automation I tend to look at opportunity first. For example how many markets are available & how often can the trade be triggered - clearly a big difference between a strategy that trades 90% of markets vs one that that may just hits a few. The later would take forever to gather enough data to know if it was really worthwhile taking forward.

I don't worry to much about scalability as who knows what the future holds and you are always learning as you go and may find other opportunities as the strategy gets developed.

But a good indicator tends to be money typically waiting in ladder when your strategy triggers..

[napshnap]

Yeah, money on the ladder near your entry point is acceptable approximate indicator, at least you'll know that your 100 pounds wont hurt your final price (wont make it less than the price that you consider a value) as much as 10 000 can.

[Euler]

Yeah, money on the ladder near your entry point is acceptable approximate indicator, at least you'll know that your 100 pounds wont hurt your final price (wont make it less than the price that you consider a value) as much as 10 000 can.

I monitor available money to measure how scalable something is. I'm more concerned with exiting a wrong position, than getting in on something as the latter often seems much easier.

I've found loads of strategies that work in the hundreds, but it gets really hard to scale into the thousands without affecting your entry or exit.

[firlandsfarm]

OK, interesting ... so far money available (the future) seems more popular than volume traded (the past). Maybe a combination of volume traded over the previous X minutes/seconds and money waiting to be matched at the instance.

[Euler]

Depending on which sport you are active on the prior volume matched can be useful or not.

In racing so much money arrives late on that most recent matched is more important.

[ODPaul82]

Tennis there's generally no inclination, you tend to be bounded by the level of the tournament (i.e. ITF, challenger, ATP/WTA 250/500/1000, Open), what round it is, time of year, etc.
December I could with relative ease put in £500 into a challenger 125 level, May/June time I'd have to be splitting that up into multiple bets.

[LinusP]

One way to slice this is to look at signal decay, this is quantifying how time sensitive your signal is vs profit. A signal in which the decay is fast will result in a strategy that is very sensitive to market capacity compared to something which tails off slowly.

IMG_2011.jpeg

[Anbell]

For automation I tend to look at opportunity first. For example how many markets are available & how often can the trade be triggered - clearly a big difference between a strategy that trades 90% of markets vs one that that may just hits a few. The later would take forever to gather enough data to know if it was really worthwhile taking forward.

I don't worry to much about scalability as who knows what the future holds and you are always learning as you go and may find other opportunities as the strategy gets developed.

But a good indicator tends to be money typically waiting in ladder when your strategy triggers..

I think this is the correct take. The number of markets and the number of trades is usually a far bigger driver than absolute size per trade.

[firlandsfarm]

One way to slice this is to look at signal decay, this is quantifying how time sensitive your signal is vs profit. A signal in which the decay is fast will result in a strategy that is very sensitive to market capacity compared to something which tails off slowly.


IMG_2011.jpeg

Interesting point LinusP, I'll look into that ... Thanks for raising it

[firlandsfarm]

For automation I tend to look at opportunity first. For example how many markets are available & how often can the trade be triggered - clearly a big difference between a strategy that trades 90% of markets vs one that that may just hits a few. The later would take forever to gather enough data to know if it was really worthwhile taking forward.

I don't worry to much about scalability as who knows what the future holds and you are always learning as you go and may find other opportunities as the strategy gets developed.

But a good indicator tends to be money typically waiting in ladder when your strategy triggers..

I think this is the correct take. The number of markets and the number of trades is usually a far bigger driver than absolute size per trade.

OK, I get that Anbell but I was really coming from this from the direction of by how much can you increase your stake and still have the edge. For example, we talk of low stakes for new strategies but when you are happy with the strategy and want to stick your neck out a bit more how do you know how far to go or is it just a matter of slowly increasing and monitoring to spot at what point the efficiency begins to tail off. A bit like the economics formula of price x number sold. You put your price up and sell less but the total turnover/profit may be higher if the decline in sales is less than the price hike!

[Bobajob]

Suck it and see

[The Silk Run]

Bank on it, go in hard, and crawl over the broken glass ...