Vladyslav Zagorodniuk / Perfil
- Informações
6+ anos
experiência
|
0
produtos
|
0
versão demo
|
0
trabalhos
|
0
sinais
|
0
assinantes
|
![Vladyslav Zagorodniuk](https://c.mql5.com/avatar/avatar_na2.png)
Vladyslav Zagorodniuk
Here is interesting article sent by Mr. Boris Schlossberg I’d like to share with you:
Algos Always Fail
After trading for so many years there are only two absolute truths that I believe in. One - all trade strategies can be reduced to continuation and mean reversion. Two - HFT aside, there is no automated trading strategy ever created that will not blow up your account in the end.
The two truths are actually inextricably intertwined with one another. The first truth basically states that markets are always in one regime or another and the intensity and frequency of that regime will determine the success or failure of the second truth. Why? Because all automated strategies are formulaic and formulas are always based on some underlying implicit assumption that we can never prove.
One very simple example of this was the orange juice market. For many years in the 1970s and 1980s, orange juice prices used to rally in fall and winter (which is peak growing season in Florida) and then fall in spring and summer. Now if you studied all the data, backtested all your spreads, and ran the most rigorous statistical analysis tools on your strategy you would be extremely confident that you could have made money with that seasonality trade.
And then of course when people tried it they lost ever penny.
Why?
Because the data only told you about the past and was utterly useless - in fact massively injurious - in its assumptions about the future. What happened in the future was that Brazil became a major player in the orange juice market in the 1990s and Brazil, of course, is in the Southern hemisphere so the trading patterns not only changed but flipped around completely wiping out every algo trader stubborn enough to trade his system for the decade.
The algos were victims of regime change as they always are and such regime changes happen all the time in the markets. These days, I must admit to no small amount of Schadenfreude as I watch all the great titans of the hedge world twist in the wind as their oh-so-vaunted value models get shredded to bits by the market that only cares about theme and momentum. (Value is inherently a mean reversion strategy while momentum is a pure continuation trade).
Now it’s bad enough to hold on to an outdated ideology that is completely ineffective in the present market environment, but it is doubly sinful to entrust that ideology to an algo that will happily trade your account down to zero as it is just programmed to mindlessly follow the orders of its formula.
Don’t get me wrong. I am not anti-algo. In fact, I spend all my free time creating formulas for my programmers so that I can exploit whatever edges I see in the market. But after making and trading numerous automatic strategies I have come to appreciate their glaring limitations - namely that no matter how clever, how selective, how thorough you think you are with your code you will never be able to make an automated strategy consistently profitable.
That doesn’t mean I am abandoning automated trading - quite the opposite - I plan on increasing their use both for my FX and my stock index futures accounts. I am, however, going to make a key change in how all my algos operate. I call it semi-auto. Instead of my algos automatically taking every trade, they will instead serve up the trade idea for me to accept or reject. The buy/sell decision will be left up to me.
I can hear all your objections now. You will miss many trades. Your own ability to analyze the market is by no means guaranteed to be better than the algos. You will emotionally crumble when your “second guess” ideas fail.
Yes, yes and yes - and still I think its better. Yes, I will miss many trades because the algo is always on and I can only pay attention to the market for a limited amount of time, but I will also miss a lot of BAD trades because the algo does not discriminate and I will. Will that result in me making a bunch of bad decisions because I am too scared to act or too eager to trade? Perhaps. But there is no running away from the fact that if you want to trade the markets well you need to learn to understand them and you need to learn to control yourself. You will emotionally crumble regardless of whether you lose money through your own mistakes or the algo does it for you. Your chance of turning off the algo at the maximum point of its drawdown is nearly 100%.
That’s why no matter how much we want the machines to do all the work for us they never will. The markets which are perhaps the single greatest example of constant human change will always require us to use our wits and our will to succeed. The machines can only help, but they can’t replace us.
Happy Trading.
B
BKForex LLC | The Desks of Boris Schlossberg and Kathy Lien, NY, NY 10024
Update Profile | About Constant Contact
Sent by contact@bkforex.com
Algos Always Fail
After trading for so many years there are only two absolute truths that I believe in. One - all trade strategies can be reduced to continuation and mean reversion. Two - HFT aside, there is no automated trading strategy ever created that will not blow up your account in the end.
The two truths are actually inextricably intertwined with one another. The first truth basically states that markets are always in one regime or another and the intensity and frequency of that regime will determine the success or failure of the second truth. Why? Because all automated strategies are formulaic and formulas are always based on some underlying implicit assumption that we can never prove.
One very simple example of this was the orange juice market. For many years in the 1970s and 1980s, orange juice prices used to rally in fall and winter (which is peak growing season in Florida) and then fall in spring and summer. Now if you studied all the data, backtested all your spreads, and ran the most rigorous statistical analysis tools on your strategy you would be extremely confident that you could have made money with that seasonality trade.
And then of course when people tried it they lost ever penny.
Why?
Because the data only told you about the past and was utterly useless - in fact massively injurious - in its assumptions about the future. What happened in the future was that Brazil became a major player in the orange juice market in the 1990s and Brazil, of course, is in the Southern hemisphere so the trading patterns not only changed but flipped around completely wiping out every algo trader stubborn enough to trade his system for the decade.
The algos were victims of regime change as they always are and such regime changes happen all the time in the markets. These days, I must admit to no small amount of Schadenfreude as I watch all the great titans of the hedge world twist in the wind as their oh-so-vaunted value models get shredded to bits by the market that only cares about theme and momentum. (Value is inherently a mean reversion strategy while momentum is a pure continuation trade).
Now it’s bad enough to hold on to an outdated ideology that is completely ineffective in the present market environment, but it is doubly sinful to entrust that ideology to an algo that will happily trade your account down to zero as it is just programmed to mindlessly follow the orders of its formula.
Don’t get me wrong. I am not anti-algo. In fact, I spend all my free time creating formulas for my programmers so that I can exploit whatever edges I see in the market. But after making and trading numerous automatic strategies I have come to appreciate their glaring limitations - namely that no matter how clever, how selective, how thorough you think you are with your code you will never be able to make an automated strategy consistently profitable.
That doesn’t mean I am abandoning automated trading - quite the opposite - I plan on increasing their use both for my FX and my stock index futures accounts. I am, however, going to make a key change in how all my algos operate. I call it semi-auto. Instead of my algos automatically taking every trade, they will instead serve up the trade idea for me to accept or reject. The buy/sell decision will be left up to me.
I can hear all your objections now. You will miss many trades. Your own ability to analyze the market is by no means guaranteed to be better than the algos. You will emotionally crumble when your “second guess” ideas fail.
Yes, yes and yes - and still I think its better. Yes, I will miss many trades because the algo is always on and I can only pay attention to the market for a limited amount of time, but I will also miss a lot of BAD trades because the algo does not discriminate and I will. Will that result in me making a bunch of bad decisions because I am too scared to act or too eager to trade? Perhaps. But there is no running away from the fact that if you want to trade the markets well you need to learn to understand them and you need to learn to control yourself. You will emotionally crumble regardless of whether you lose money through your own mistakes or the algo does it for you. Your chance of turning off the algo at the maximum point of its drawdown is nearly 100%.
That’s why no matter how much we want the machines to do all the work for us they never will. The markets which are perhaps the single greatest example of constant human change will always require us to use our wits and our will to succeed. The machines can only help, but they can’t replace us.
Happy Trading.
B
BKForex LLC | The Desks of Boris Schlossberg and Kathy Lien, NY, NY 10024
Update Profile | About Constant Contact
Sent by contact@bkforex.com
: