Averaging unprofitable spreads.

Посмотрел вчера video на смартлабе. Turns out, averaging to infinity with a fixed step, without stops, now called a market making strategy. I have already spoken about this that, really, thus, you can be afloat for a long time until a long tail appears. Это случается, at least, every few years. But before that tail, в течении длительного времени,  you can receive management fees and percentage for success. So it's quite a profitable strategy for managers.. We don't take investors into account.

But here comes another point. Type, consider that if we have on the MICEX 100 listed companies, then creating new instruments from them in the form of spreads, получим 5000 new instruments, independent of the direction of the market, which we will trade against the trend, averaging to infinity and someday we will still come out in plus. That is the risks, ostensibly, zero. The more tools we use, the more chances of being in profit.

So I understand, the bet is on, that new instruments in the form of spreads should have one feature — counter-trend systems must be profitable for them. Otherwise nothing will work. This is the cornerstone. Why on earth should there be a statistical advantage of counter-trend systems for stock spreads?? Можно очень просто убедиться что charts новопостроенных инструментов ничем не отличаются от графиков обыкновенных акций — they have exactly the same trends and the same flats, periods of high volatility alternate with periods of low volatility. Remove the vertical scale and it will be impossible to tell where the spread chart is, and where is the graph of a single instrument.

For some reason, adherents of pair trading argue that the spreads of past correlated stocks should converge after the divergence. I personally, I don't see on the charts that they always converge with a profit. In the sense of returning to the moving average. Very often returns to the average already at the deepest loss.

  Investidea: Jack Henry & Associates, because they counted all the money

And again, there is a sliding window problem, за период которого измеряется correlation tools. If, for example, take for a year, get one value, and in six months or two years, completely different and may differ several times, и за один период может быть хорошая correlation, and for another period, at all, uncorrelated.

That is, all scientific reasoning about market neutrality, minimized risk, market making, etc.. worth nothing until it is proven that stock spreads have a statistical advantage in trading counter-trend systems. In what, лично я, I doubt very much.

Scroll to Top