April 8, 2016

There are millions of traders and millions of trading methods, but a lot more investors. At the end of the day, all financial assets are accounted for to the penny and in someone's hands. In the US, that's about $99 Trillion dollars worth; this includes real estate, stocks, and bonds. It's a big number. Some hold some of these assets for a short time, others up to multiple decades.

The short-term retail trader is part of the minority and doesn't control anything. 

If you think they do, please cite a study or academic paper that could corroborate your claim. The short-term trader should be considered an in-and-out guy using whatever reason appeals to him or her. 

From what I read in papers, the media, and trading-oriented forums, the short-term trader appears to not be that proficient in statistics, fundamentals, math, or probabilities. He just makes theories up as he goes along without much in numbers as supporting evidence. You will read claims like: my "system" has an 80% hit rate; with 4 of the 5 discretionary trades made being profitable. And based on that flimsy claim one should pour capital his way to manage. I know I exaggerated a bit, but everyone should get the message. 

I think the short-term trader has a propensity to simply gamble on almost the same basis as a casino player. Talk to him about odds, probabilities, win/loss ratios, risk/reward ratios, long-term outcomes, or whatever. He has a short-term trading system, will make up the numbers to justify his stance, and will stand ready to sell you his system should you ever want to agree to his no-refund policy. After all, you would have seen his "secrets". 

You can't criticize their ethereal "secret sauce". They will tell you that you are too feeble-minded to really understand all the intricate and complex stuff they put in their black box to achieve those outstanding backtest results. You question the math behind their endeavor, and they will reply with: what math? It's all math; this thing follows my "proprietary" mathematical indicators, which can forecast future price levels. You will probably get some excuses when asking to show any kind of long-term backtests or simulations. Something, anything to show that it could at least have happened in the past over extended periods of time. 

What I see most often are what I call peanut discretionary players making short-term discretionary peanut plays, hoping for peanut rewards, which they don't always get. A lot of them would like to manage other people's money with trading strategies that are so poorly designed that they can only blow up with time. In a lot of cases, you won't have to wait long to see their strategies fail. 

They could have designed their trading systems better, of course, but they did not, and by choice. They willingly decided to accept these flawed strategy designs. And then, they would like to pass them on "for a fee". It is really a case of caveat emptor. 

Why would I say such things? When you see trading strategy designers worried about commission costs, slippage, equipment failure, internet downtime, or whatever insignificant trading problem, then you have to conclude that if peanuts matter so much, it is because it has become a major part of their potential profits. And since commissions are less than 1% and downtime is less than 0.001%, they certainly weren't aiming very high in the long-term CAGR department. 

But who am I to even criticize their methods? Whatever. They almost never display summary performance reports which are generated by their trading software each time they do a simulation. If they did, at least they would have some statistical evidence, some ballpark figures to show. I know I can read these reports with ease. So, why don't strategy developers not show their long-term summary performance reports? The reason should be evident. 

Some will categorically state that all trading strategies break down with time. You don't abandon a trading strategy that is making you money unless it is not enough. You abandon losing strategies, non-productive ones, while the good trading strategies might be replaced, but only with even better strategies. 

The only reason for someone accepting that a strategy has failed is because it has lost a lot more money than their original author thought even possible. It is the same reasoning for always tweaking a strategy on a monthly basis or whatever. It is because it did not perform as expected, whatever the strategy might have been. You have to cut losses before you bankrupt your trading account since after is not the best of decision points, the damage will have been done. 

If you have a peanut account, forget about HFT. As a matter of fact, you should probably forget about playing this game altogether. 

Do the math. Whatever your stake, you will have to abide by A(t) = A(0)*(1+CAGR)^t. There is no escape from this. So if your stake is small, please consider the number of years it will take to reach whatever A(t) level you aspire to. Then consider the time you will have to spend on the problem, first developing something worthwhile, then monitoring your trading program for the years to come. You might find that on an hourly basis, it was not only boring but that you were playing for peanuts. 

Then again, some will say that peanuts are good, even when about anything else might have produced more. 

I say that whatever trading strategy you may have if it is profitable, you must continuously go for more capital. Otherwise, you are really shooting yourself in the foot. Look again at the expression above; it is rather explicit. For me, anybody with a profitable trading strategy seeking additional funds to manage is more than justified. I would consider them lacking in a big way if they did not. I would classify it as a not-so-bright move with the consequence of not hearing whatever they had to say since about anything could outperform them, even buying index funds. Let it be said, if your automated stock trading strategy cannot outperform a long-term index fund, then it is worth shit. Buying those index funds would have been a better decision. 

An automated stock trading program is just that, a software program. It should not care about the size of the stake to be managed or what it will be. But if your strategy is only able to manage peanuts, why should I even care about your CAGR? Your strategy has no real long-term future. Not because it cannot make money, I'm sure it might, but because it is not making enough. Not enough so as to outperform the averages or index funds. 

Anybody, and I do say anybody, who can program can program a worthwhile trading strategy. How hard can it be? You buy some stuff, resell it later, and make a profit or not. What is there not to understand? 

Here is the basic code framework, a single do-while loop till not last bar: 

var Bar: integer;
for Bar := 20 to BarCount - 1 do // from first tradable bar to last bar
  if not LastPositionActive then
   { Entry Rules }
    { Conditional Entry Code }


  { Exit Rules }
   { Conditional Exit Code }

If you want a sure thing, forget it. But if it was the case, and you had one, you would not need a trading strategy in the first place; you would simply execute. 

You develop a trading strategy because you don't have a sure thing, and you know it. 

You are ready to play averages where, hopefully, you will be winning more than you lose. That is why you do those backtests: to gain confidence in your trading methodology. Now, if you don't do your job properly, cut corners, curve fit, peek (cheat), don't expect me to follow in your footsteps either. Should you put out performance results that are statistically insignificant or cherry-picked examples, again, forget about it. 

I cannot reject or deny all the studies and academic papers that have been written over the past century. On the contrary, I rely on them to explore trading avenues that have been considered in the past and test them with programs and state-of-the-art equipment. I will also modify them to suit my current views of the problem as if always seeking better ways. 

But before accepting any new programs, they will have to perform better than what I already have, otherwise again why bother, they simply end up in my code snippet library as a nice try but no candy. At least, I would know another way of not doing it and might have more reusable parts and software routines that might serve in other programs. 

In the end, you will find that you have only one person to convince, and that is you, nobody else, just you since you will be the one to monitor your trading programs. They better do what you programmed them to do. It is with that conviction that you will be able to interest others in your trading methods, if at all.

Created... April 8, 2016,    © Guy R. Fleury. All rights reserved.