just a comment regarding the May 6 data. It might have been a black swan.. - but these things happen. So, it is good to cross-check whether any strategy you are dealing is not burning your money in there.
The way I handle this: I keep weeks of data and typically do cross- checking on per week data (as well as optimization). I do not use this week data (I actually exclude week of 16.5 as well) from any optimization or cross-checking when I do an initial assessment. When I then find a strategy I check them with these two weeks of data as extreme cases. My experience is that the flash crash leads to exteme over-estimation of returns. Basically all strategies I tested would have generated wind-fall profits in there. (Of course this is theory, I do think the spreads were rather high and movement was very fast, so you never know where the market orders would be actually filled.) On the other hand week of 16.5 was for all my otherwise good strategies very bad (has anyone a strategy which works in general and for this week?) So, these are two movers that will distort the picture whenever you keep them in optimization or regular cross-validation. (But you should still keep the data and use it as a final test. - At least that is what I am doing. And the final question to myself is always: can I stand it if something like this happens - which is of course mostly the issue with May 16-week.) Just my 2cent Klaus On 10 Jun., 04:21, Keith <[email protected]> wrote: > Very interesting guys, many thanks for the great explanation. > > I have observed the same issues you rightly noted and as a result I > only look at price action - which is the outcome of any trading > decision. > > Is there a strategy that looks at price action, then use depth balance > to support the direction. > > With our naked eyes, we can sometimes see a sudden push to the upside > or to the downside - is there a way to detect the first one or two > ticks of the move and join the party as early as possible? > > On Jun 9, 2:40 pm, nonlinear5 <[email protected]> wrote: > > > > > > Depth balance is a number between 100 and -100. At one extreme, all the > > > orders would be on the bid side, on the other extreme they are all on the > > > ask side, and at 0 there are equal numbers of bids and asks (adding each > > > of > > > 10 levels on the bid side, and the ask side). > > > Just to expand on this a little. The idea with tracking depth balance > > and its changes (known as velocity of balance in JBT) is that the > > Level2 sizes and prices represent the current demand and supply > > (buyers and sellers) in a particular market. Depth balance in JBT > > measures exactly that. Of course, if things were as pure and linear as > > the classical economics prescribes it, your strategy would buy when > > the demand exceeds the supply, and sell short otherwise. In reality, > > things are much more complicated. > > > First, there is a lot of noise which obscures the patterns. > > > Second, there is a lot of game playing in the L2 where players flash > > bids with the real intention to sell, and flash offers with the real > > intention to buy. > > > Third, there is a powerful factor which works against the classical > > theory. This factor states that the market will move to the point of > > the highest liquidity. That is, the price will move in the direction > > where the highest number of shares/contracts can be traded. If you > > think about this, it makes sense. The large players moving a lot of > > money in the market don't care where the price moves. They only care > > how they can make the largest amount of money with as least risk as > > possible. They will go long as easily as they would go short, > > especially in the futures markets. So, let's consider a classical > > case. The demand is 10000 (that is, the cumulative bid is 10,000 > > contracts), the supply is 1000 (that is, the cumulative offer is 1,000 > > contracts). Clearly, the demand far exceeds the supply, so by all the > > scientific economic principles, the price should go up. Now imagine > > that you are a big guy, and the way that you make money is move 10,000 > > of contracts at a time. In this particular situation, there is not > > enough liquidity on the supply side to move your size, but enough on > > the demand side. That is to say, it's easier for you to actually sell > > 10,000 than to buy 10,000 contracts. As a result, while everything is > > telling us that the price is about to go up, it may in fact move down, > > as a result of you "selling into liquidity". > > > So, all of these factors complicate things. Ultimately, it comes down > > to whether you can somehow take them into account and come up with a > > strategy to take advantage of them. -- You received this message because you are subscribed to the Google Groups "JBookTrader" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/jbooktrader?hl=en.
