risk management for day trading

Quote from Yannis:

Some time ago, one of my teachers demonstrated to me that if you just have random entries (e.g., regular coin flip) you can still be profitable if you define the exit strategy and implementation well. And if you do that very well, you can be very profitable. Profit is made upon exit, not entry. Couple that with a decent way to manage the amount risked and the profit targeted per trade, and you have a solid system. That's not what I do - entries for me are very important (if nothing else, for psychological reasons) but I do know that his math is correct. In other words, he showed me that mm can be a sufficient edge on its own.
More...

This may be the sort of thing that is theoretically true, but it's often surprising to new traders who believe it how much money can be lost by "proper management" of random entries. It's not the sort of thing I want to argue about anymore since all I can do is repeat what I've already said. But, in practical application, I find the entry to be every bit as important as the management of the trade.

As for defining "money management", you are correct. However, there must first be money to manage, and there is a considerable difference as to what the theoretical and the practical can accomplish.
 
problem is :
markets aren't random.

also, there is a big difference between a general probability of an outcome (among many trades) and the local probability of each trade.

ie try to get long in a strong down trend.
you can try 10 times, 20 times, 30 times, based on your exit / stop loss parameters, your probability can be very small to get a profit, while your system, 'in general' might have a win ratio of whatever (30%, 60%, does not matter).

so yes, entry is very important. or more precisely the side you choose. even if you are a non directional trader as I am, direction is still a component in the trade.
the entry level can be optimized or not, that's another story (scaling or not). but the key is the direction bias of each trade.

I take my example from above, but reverse it.
now the trend is still down, but this time I short, 5 times, 10, 20 times. and I make money 100% of the time (let's say, once again that stops are far enough to avoid shake outs).

system may still have the same long term probability. but the LOCAL probability was very high, albeit temporary.

anyway, there is hardly another way than to use long term probabilities (if possible actual ones as opposed to backtest) and apply some position sizing to it.

in futures, you need to deal with the risk of ruin as well. an extra very difficult parameter to include in your model.

still, not much can save a negative expectancy.
all this risk management is a way to OPTIMIZE a positive expectancy system (or eventually a break even one, but that's not as strong a system then).

anyway. good luck
 
When I think of money management I think of "position sizing". or how many shares/contracts for the next trade based on the size of your equity. Various methods to determine this, a fixed fraction, Fixed ratio, set percentage etc....

But if you aren't going to be profitable with 1 contract(or 100 shares) ie. the expectation of your trading over the life of all trades, is negative; then position sizing can't make you profitable. However, it could possibly give you exponential equity growth (or the flip side it could make you go broke) with a positive expectation.
 
Quote from tntneo:

ie try to get long in a strong down trend.
you can try 10 times, 20 times, 30 times, based on your exit / stop loss parameters, your probability can be very small to get a profit, while your system, 'in general' might have a win ratio of whatever (30%, 60%, does not matter).
so yes, entry is very important. or more precisely the side you choose. even if you are a non directional trader as I am, direction is still a component in the trade.
More...
tntneo,

Actually, yours is a good example to prove my point: that one can be profitable with random entries and carefully managed exits. You are talking about strong trends - no problemo, imagine the following, very approximate, system:
1. ADX >30 or 35 etc DMI+/- whatever parameters you want, make sure you trade in a strong trend;
2. Randon entry (ie, 50% long and 50% short in the long run, etc)
3. ProfitTarget = 2*StopLoss > average pullback

Very rudimentary but it'll work. One can easily expand with multiple lots that exit at different places (e.g., exit the first at a decent place, bring the rest to B/E, use an advanced (eg, chandelier) stop strategy afterwards, etc etc

Not that I trade this system, but it can be used to show that random entries coupled with carafully managed exits CAN make a profitable system.

As I remember from that demonstration, the opposite is not as easily shown to be true: carefully planned entries and random exits have a harder time staying profitable, unless you use multi-level entries and single, random exits, which most traders would find counterintuitive. Still, the system will suffer because you want to put the elaborate management (in my first example, managing the exits) at the point in time when you have the maximum information about the local market conditions. In other words, do the random thing first and then use the additional information you gather to guide you in your exit strategy.

Make sense?
 
First, you're not making random entries. Second, why must it be a choice between random entry and carefully-managed exits, and carefully-managed entries and random exits?

If one is going to make random entries, he must be prepared for an unbroken string of losses. Each loss reduces the account. If the same percentage is applied to each trade, it becomes next to impossible for the trader ever to get back to breakeven.

Those who continue to insist that random entries can be profitable rely on "studies" or something they read somewhere or something that somebody demonstrated in class. The practical application is a different story. Until somebody can produce an actual system of random entries that, when applied, yields a greater profit than the same system with non-random entries, I'll continue to go with the lessons of experience.
 
Quote from dbphoenix:

First, you're not making random entries. Second, why must it be a choice between random entry and carefully-managed exits, and carefully-managed entries and random exits?

If one is going to make random entries, he must be prepared for an unbroken string of losses. Each loss reduces the account. If the same percentage is applied to each trade, it becomes next to impossible for the trader ever to get back to breakeven.

Those who continue to insist that random entries can be profitable rely on "studies" or something they read somewhere or something that somebody demonstrated in class. The practical application is a different story. Until somebody can produce an actual system of random entries that, when applied, yields a greater profit than the same system with non-random entries, I'll continue to go with the lessons of experience.
More...
db,

For the sake of this example, I did define the entries to be random - just a regular coin flip, I said, meaning that you go long or short randomly. You can also do this by adding random times to enter, same thing. All I was trying to say is that random entries and carefully managed exits CAN make a profitable system.

As far as the value of theory vs experiment goes, it's an old argument, and I've lived through it intensely (I have a PhD in Theoretical Physics.) The idea is that theory allows one to scan though many possibilities rapidly and target the one or two juicy situations to experiment with in depth. The alternative, experimenting with all possibilities, is much less feasible.

What this teacher taught me is that exits are more important than entries. (This was a trading system designing workshop.) Therefore, as I design trading systems (which I like to do) I spend a good amount of time designing the entry and twice as much designing the exit management strategy. That's all. Actually, as I said in a previous post, (generalized) money management is very important to me. And, many of my systems trade in real life quite well.
 
We're getting circular here, but, again, claiming that random entries and carefully managed exits CAN make a profitable system is not the same as actually producing one, preferably one that produces more than a subsistence income.

And, again, what some teacher said or what one learned in some class may not necessarily apply to practical use.

If a beginner wants to believe that he can enter wherever he likes as long as his exit is pre-planned, he should be prepared for at least the possibility of a string of losses that may do considerable damage to his account.
 
Yannis, I did not back test your system.
but we are actually saying the same thing..

look,
your entries are not random, they are filtered to find a trend (!).
within a trend, you can use random entries that's true. my point was not to show you can't use random entries, but to prove the market is not random, therefore within one established direction you can pick any level and make money (unless you screw up the exits).

so we agree.
but again, I don't call these random entries !

a random entry is when you flip a coin every hour or day or 15mn whatever and place a trade.
apply the best exit strategies you want. it won't work.

tntneo
 
tnt,

I have to look back and find my notes to make sure, but I remember those discussions and say that even what you describe is possible.

Let me clarify:
1. System enters at random moments in time and in random direction (long or short);
2. System does not enter again until position is exited;
3. Exit strategy takes it from there.

Question: Can one design an exit strategy that makes the system profitable over reasonable periods of time?

Answer: I believe so.

The idea is that one has to load all market intelligence (eg, current direction and stregth of trend, volume, info from leading indicators, etc, etc) into the exit strategy, and execute. And because there's more information available as the trade evolves, that side (the exit side) is the superior place for that intelligence.

How easy is that? Not very... it's a task for experienced system designers.

How can one tell upfront that this is not a fruitless effort? Remember, I believe that it is reasonable to assume that if that system's exit was also random, the system would only incur the cost of slippage and commissions in the long run. So, what the "intelligent" exit strategy has to do is counterbalance those costs, at a minimum, which, imo, is feasible.

Have I done it? I have designed several systems (and traded a couple for real) that have a lot more smarts on the exit side as compared to the entry. For example, systems that take in a relatively simple algorithm based on pivots, H/L etc and enter a minute or so after open. Then they think long and hard as to where to put stop loss levels and profit targets and how to manage those. Not the same but similar. I've compared mine to systems that think long and hard where/how to enter and always exit at the end of day. Mine are better :)

Again, as we all agree, not a task for a beginner.
 
You must log in or register to reply here.
Back
Top