Re : Does Scalping work anymore

[LeTiss]

Just been chatting with Betfair support on Twitter.

It seems that, because I made a net loss during the hour in question, I'll be fine (though the adviser wasn't very clear).

He told me: 'This is the activity on Chilli Jam'. Yeah, cheers mate, that clarifies it perfectly!

Jeff

In other words, you don't have enough money in your account that's worth taking

[Iron]

I revisited the scalping today, this time focusing on the second and third favourites.

I didn't do as badly as last time, but still lost money - £4.92 loss on £1186.75 turnover (-0.27% ROT).

I was up after the first hour and quote enjoying the trading, till the 16:25 7f Nov Stks, when all my good work was undone by Agars Plough.

I've attached a spreadsheet relating to that race, with bets in ascending order by Betfair ID (i.e. the oldest at the top). As you can see, before I redded up I was 18 ticks down, wiping out my patiently accrued gains in other races.

Curiously, this experience corresponds with a post I wrote nearly six years ago -
viewtopic.php?f=17&t=5019.

Any thoughts?

Jeff

Trading Experiment 3.xls

[Euler]

I can't actually work out what your strategy is? On Agar's plough you seem to just be pumping in multiple orders? You have way more lay than back orders?

[Iron]

The strategy was for the second and third favourite, and was:

- Offer a lay to the market at the bottom of the current spread.

- If the lay is taken, offer a back on the top of the spread.

- If the price drops a tick, offer a lay at the bottom of the new spread. If the previous trade hasn't closed, offer a back bet at the top of the new spread.

- If the price rises, offer a new lay at the bottom of the new price.

- Green or red up 10 seconds from the off. Do not close any open orders prior to this point.

I appreciate that I wasn't that far from crossing the line to profitability. Any advice on tweaks I could make that might take me to the other side of that line would be appreciated.

BTW, my biggest concern when running the experiment was avoiding running up transaction charges for going over 1,000 trades an hour. Does anyone know of any way of ensuring that I don't make more than 1,000 trades an hour, apart from continually monitoring my bet history?

Thanks

Jeff

[foxwood]

Idle thoughts on transactions ...

Worst case races per hour at 5 minute intervals assuming no collisions = 12

Max bets per race = 1000 / 12 = 83

Max trades per race assuming simple place/offset each time = 83 / 2 = 41

So, on a busy day the most trades per race would be 40

With 80 bets in 5 minutes that averages 7.5 secs per trade - unless you are looking for micro signals in price movements you don't need a fast refresh rate.

Adjust all of that for placing an opening offer then cancelling if not matched in x seconds - each bet counts whether matched / cancelled / lapsed.

Analyze your strategy in depth to work out the real details of bets to be taken / offered bearing in mind the overall limits.

[xitian]

I think with scalping you need to do more than just "earn the spread". Well that might work in some markets, but probably not in very efficient markets like horse racing.

The instances where you do make a successful trade will be offset by the ones where you've offered up a "bad value" price and someone else with more information than you takes it for a profit themselves (at a loss to you). So basically, you do need to have some form of edge or signal to tell you which way a price is more likely to move. In the same vein to the above, [your latency in] cancelling bets is just as important as placing them. You need to remove your offers when your "signal/edge" tells you that the price you're offering is no longer good value to you - for example in the case of a massive spike against you, well you should have known to cancel your offer before that spike was likely to happen.

Now you could do this at the macro scale and say "don't scalp races of this grade and X number of runners", but you can also develop more micro rules and trade all races and instead not offer prices at every single instance as I've mentioned above. Only offer in favourable circumstances on the favourable side. How you define "favourable" is where the (extensive) work of finding a small edge comes in, but that's essentially trial and error and some imagination.

I also find that being at the front of the queue (ie. offering into empty prices) is generally an advantage, if it's a sensible price of course. After all, you can cancel it whenever you like at no cost if it becomes unfavourable, but while it's favourable, you're first in line to be matched.

[Iron]

Thanks Xitian

This scalping lark isn't straightforward, is it?

I'm sure I could find a scalping edge if I were to play about with enough parameters, but (big meetings aside) I get the impression that there are more profitable approaches.

You make a valid point about offering 'out of money'. I think the challenge, however, is that you're opposing the trend by so doing (if you offer to back at 6 when the price is at 5, for example). I haven't been able to make out of money offers work, but I keep an open mind.

Jeff

[spreadbetting]

I'm sure I could find a scalping edge if I were to play about with enough parameters, but (big meetings aside) I get the impression that there are more profitable approaches.

In general the only thing scalping is good for these days is selling ebooks and pulling in punters to training courses

Scalping randomly gives plenty of wins so it's more than enough to convince people they can crack it but it's certainly no longer as simple as it appears and there are more profitable, and easier, ways to make a buck on Betfair. I'd look towards those Jeff as I've never thought botting is really suited to scalping.

[PeterLe]

Id probably agree with that (I dont tend to scalp either these days).
I just found the file that I used in July last year; Id be happy to run it again tomorrow using small stakes (£5) and post the results if anyone is interested?
Regards
Peter

[ruthlessimon]

Id be happy to run it again tomorrow using small stakes (£5) and post the results if anyone is interested?
Regards
Peter

Yes please Pete, I'd be very interested

[Crazyskier]

Id probably agree with that (I dont tend to scalp either these days).
I just found the file that I used in July last year; Id be happy to run it again tomorrow using small stakes (£5) and post the results if anyone is interested?
Regards
Peter

Yes please

[PeterLe]

Sure no problem; I’ll run it on 5 stakes with a hard stop if it loses £20

[PeterLe]

Well here's the results from today

Running entirley on auto (excel/Guardian)
Traded 12 markets

Tot amount of bets = £1877
P&L -7.96

The big losses were due to the horse steaming towards the end coupled with lower SP's than anticipated
Im sure there are some people out there who can make this pay, but I dont like the risk/reward, so back to my normal BOT for the Kempton meeting; Im out!

Regards
Peter
PS One point worth mentioning; I wouldnt normally scalp anything with odds of less than 3 but I did today. If you remove those trades it comes in a bit better.

[Euler]

There were loads of short-priced steamers around today.

[Iron]

That's interesting Peter, thanks for running the test.

When I returned similar losses the other day, I wondered if the fault was with my method, as I would have expected far lower losses (or maybe even a small profit).

It would appear, however, that the market has changed since your and Euler's previous tests, and that perhaps you need to do more these days to get an edge than merely offer to either side of the book.

Jeff