Hello,Has anybody an idea how the execution speed may affect the profit?I think, that at each instance a number of arbitrage possibilities are created, N(0). If to wait t_delay number of second these possibilities will be taken off at the rate of exp(-t*a) (a-some constant).Another factor to consider is the size of the market. There the number of arbitrages grows quadratically with number of securities one monitor,Then, we come to estimation of PnL as a fanction of reaction time and size of the market we monitor:PnL(t_delay,N_instr) ~ N_instr^2 * exp(-t_delay*a) , a - decay constantHas anybody made any estimations of speed effect?Nik Text

we are talking of course of risky arbitrages,I think that this speed is function of the risk and more speicifcally of the sharp ratio specific to the arbitrage.Of course most ofthe time it will be relatively easy to specify the return (at least stochastic), but the risk will be always very difficult to estimate due to the all scale of market risks, credit risks, operational risks, model risks,..

I assume that when we enter the position, we have some estimation of risks and profitability.of course they are stochastic.I m thinking of different thing: the profitability of the trading system as a function of its reaction spead.Let's take ADR market with static arb. Once I see arbitrage I am willing to trade. However, there are many people on the market willing the same.If I have super-sonic trading system then I am the first in the queue always!!!More difficult situation is for dynamic arbitrage, i.e. when we trade NOW vs FUTURE. Stochastics enters here. And even more difficult if it involves option pricing. Change of underlying price induces change of option price. Fastest pricing system gives obvious advantage....Nik

nikol, of course fastest is best. on eurex for example, there are often cherries that hit the option marketat a price everyone wants to trade. having the fastest machine guarantees that you will get theseprofits.the tough question is how much is this worth, because fastest is also expensive...

[blockquote]QuoteOriginally posted by: filthythe tough question is how much is this worth, because fastest is also expensive...[/blockquote]this is what i try to figure out from general principles. before investing money one has to realise projected profitnik

you also have to bear in mind that the market dynamics can change very rapidly.for example, your supersonic machine picking off cheries on eurex will be renderedobsolete if the rules are changed to allow continuous mass quoting. no onewill sell you an optionat a ridiculous price if there is a sensible bid at all times.

what is it: 'continous mass quoting' ?I did not get the point. Why the system should produce 'ridiculous price'? It means that calculations/methods/assumptions are wrong. It has nothing to do with the speed.If you mean liquidity, then it is a different issue...nik

this is what i meant.say you think an option is worth 200. another decent trader may think it is worth 199 or201, but all the pros are in the same valuation ballpark. a broker needs to sell a 10 lot.he won't bother calling to get a fair price for such a small order. he will just look on bloomberg andsee the last trade was 188 (obviously a bit old so the underlying has moved). so he enters a sellorder to sell 10 at 188. this is a huge bargain and all the market makers try to buy it.the supersonic trading machine gets it.now if eurex had continuous mass quoting, market makers would need to provide updated quotesin all strikes at all times. now when the broker looks on his screen he sees a 197 bid. he hit it.your superfast machine is now irrelevant.

of course, larger my speed (and of everybody) the less is spread, the higher is liquidity, the bigger volume I can filter out, the profit is constant.probably, assuming my initial formula is correct i can derive spread at variable speed of the market?i think, that in such super-sonic market MarketMakers's will trade with each other and those who have fastest machines will pump all the money out of slow one.nik

i think the point you are missing is the importance of visibility.

In the US future exchanges there is a manual execution system called Arb desk, of course it is very fast but pretty costly. So, if one trades size he's able to do a lot of scalping and/or arbitraging in either markets.

to filfy:your example is still puzzling to me. it does not proove anything.if i give consistently not optimal price (outside best bid/ask) then by your example i get the trade, but it will not be optimal. or i got the example wrong?does ever broker look on 'last trade' rather than 'best bid/ask'? never. he always gets best price available. otherwise HE will get out of business pretty fast.it is true, that someone can have a chance to put better price than my system does, if you mean that.it is true, that my system will be irrelevant, if someone consistently puts his price better than mine. but this may happen if he has faster response (assume the same level of knowledge we put into competing algorithms).i m talking of statistically average result, not the one particular trade outcome.nik

QuoteOriginally posted by: nikol>>your example is still puzzling to me. it does not proove anything.it proves a couple of things. mainly that if there is always a tight bid and offer then speed really won't helpin a market where orders are given time priority. >does ever broker look on 'last trade' rather than 'best bid/ask'? >never. he always gets best price available. otherwise HE will get out of business pretty fast.this situation has been going on in eurex for at least eight years. there are electronic trading machinesand the faster ones do ok, but on the ise there is really no point because the markets are always there.>it is true, that someone can have a <b>chance</b> to put better price than my system does, if you mean that.>>it is true, that my system will be irrelevant, if someone <b>consistently</b> puts his price better than mine. but this may happen if he has faster response >(assume the same level of knowledge we put into competing algorithms).he doesn't have to respond. he will already be there.>i m talking of statistically average result, not the one particular trade outcome.i understand. but speed is only an advantage for getting sloppy market orders. otherwise markets generallywork on a timestamped basisnik

sorry.i completely mauled the quoting there.(although i get the feeling that there are only two of us on this thread.)

thanks anyway. your discussion helped.nik----------------------------------------------"questions are stupid, but answers are clever"