Thursday, October 25, 2012
Yesterday TMM were tripped into thinking about the complexity of algorithms and their place in this world against old fashioned humanness.
This subject quickly emerged in our chats with respect to markets, with most of TMM expressing the opinion that none of them had an intelligent thought to express today. This then led to the observation that there was little point in having an intelligent thought as this year has seen supposedly intelligent process regularly run down by algos anyway. So perhaps the non-clever clever thing to do instead, to counter the algos, would be to put on trades that are so unimaginably stupid on conventional levels they may just be stupid enough not to be sniffed out by the clever bots. "That'll fool em !"
Which led us to consider the background issue of the European Parliament and their budget process - Could it be that they've been using this method of "so stupid it may just work" for years?
But back to High Frequency models. Though TMM have found themselves run over by them in the past, we don't feel it fit to complain other than about ourselves as we are not evolving fast enough to anticipate the "thought" processes of this type of player that now sits at the poker table of markets. TMM do not think the answer is to ban them from the casino just because they win at other people's expense but to learn and adapt. TMM simplistically think that if algos are causing distortions in markets that move prices to places they really shouldn't be on your own measure of value, then you should celebrate the opportunity they give you (even in flash crashes) not curse them. If however your view is that they are terrible because once you own the position they push prices through your stop loss, then really you do have to get a grip on what markets are. If someone owns a share then they should be allowed to sell it for whatever reason they like, whether you agree with their method of analysis or not. That's just how it is. If we start stipulating the methods of investment process a fund manager has, or even day punters, then you might as well fix prices daily at a committee made up of members who have elected themselves as spokespeople to represent only their own views over those of a huge disparate seething mass of differing opinions.
Oh good Lord. No! It can't be that we have just described the EU Parliament and their budget process again, could it?
But HF algos weren't actually the trigger to this debate, it was started by the problems we humans have now convincing computers that we are humans and not other computers. Most web pages insist on the user going through some sort of test. It started fairly easily with typing in the clear letters displayed in a box, but now one has to squint through reading glasses at a blurry picture of someone's door number and then untangle a ball of string to pick out the letters of a word that isn't a word. It's a nightmare. Even persuading a bank telephone service that you are a) human and b) not a human criminal mastermind is equally challenging. Don't they realise that they are creating, through Darwinian selection, a process that will ensure that only algos will ultimately be able to pass these tests?
If computers are so so smart these days, then TMM suggest that instead of trying to prove our humanity through higher intellect we should switch tactic and instead prove our humanity through our stupidity.
"Can we have the 3rd, 75th, and 125th Cyrillic characters from the original Tolstoy novel we gave you to read when you opened this account followed by the hash key?"
" I never saw Toy Story and what the f is a hash key? Just gimme my balance you f'in stupid machine"
"Thank you we have identified you as human, we will now connect you with another human who will be able to understand your low level intellect"
" XXXX off"
Or just stick a "please use other door" sticker on the only opening door of a pair. That should keep the bots out but let most of humanity in.
Or perhaps we should just go short of usd/jpy.