Australian (ASX) Stock Market Forum

Scalping Lynas experiment

Jump to new
Sinner.
He guesses it.
Well he tries to work out from depth and a bit of tape reading
where he believes the low of the day is.
He is attempting to get it withing 3 price points (1.5C).
Some days he will get it exact.
Others within the boundaries and others plain wrong.
Its these that concern me as your waiting then until next open to get out of a bad position.

I see it as simple in its idea but far from simple in practice for the reasons I have already stated.

But no harm done its good to test an exercise like this as its how most traders come up with one.

My and your personal view (I maybe wrong) is it has limited upside because we can see the issues--as we see the issues.
 
I don't need back test to tell me if I buy close enough to the low of the day, sell at the close on an instrument that has good enough range I'd be doing pretty good.

Problem is you can't actually backtest the performance. As the key performance driver is simply your ability to pick the bottom.

Even with intraday data, the best test you could do is to find out what your performance is if you are 4, 5, 6 etc ticks away from the low. And augment that with say a 1, 2, 3 or 4 tick stop.

You can forward test your skill in picking (or getting close to) the bottom - which is basically what you are doing now. Say with practice you get it right 75% of the time. But surely by that stage you will need to add a stop component to the 25% of time where you got it wrong.

And if you can acquire the skill to find the low so efficiently, I wouldn't use that skill on LYC (which is usbjected to company specific tail-end risks - i.e. trading halt any minute). You should just take it to the various index futures...
 
skc said:
I don't need back test to tell me if I buy close enough to the low of the day, sell at the close on an instrument that has good enough range I'd be doing pretty good.

Problem is you can't actually backtest the performance. As the key performance driver is simply your ability to pick the bottom.

Even with intraday data, the best test you could do is to find out what your performance is if you are 4, 5, 6 etc ticks away from the low. And augment that with say a 1, 2, 3 or 4 tick stop.

You can forward test your skill in picking (or getting close to) the bottom - which is basically what you are doing now. Say with practice you get it right 75% of the time. But surely by that stage you will need to add a stop component to the 25% of time where you got it wrong.

And if you can acquire the skill to find the low so efficiently, I wouldn't use that skill on LYC (which is usbjected to company specific tail-end risks - i.e. trading halt any minute). You should just take it to the various index futures...
Click to expand...

skc, it helps greatly to know in advance what sort of returns you can get if you can pick the bottom. It helps to know which stock to trade this way (some it doesn't work on). And it helps to know how close to the Low you have to get to be profitable. All these things are backtestable.

When tested, they tell me:
1. the returns are huge if you can read the depth to within 3 price levels of the low on LYC.
2. the returns are pretty close to zero tending negative if you are 4 or greater levels away - in other words, if a low of 1.16 has been already printed, don't buy at 1.19, put in a buy at 1.175 and hope to get filled. Otherwise leave it for that day.

Futures would be a better instrument, as you say. This thread is about how one might trade a stock intraday.
 
tech/a said:
Sinner.
He guesses it.
Well he tries to work out from depth and a bit of tape reading
where he believes the low of the day is.
Click to expand...

Thanks tech.

My and your personal view (I maybe wrong) is it has limited upside because we can see the issues--as we see the issues.
Click to expand...

Agreed.

Ok GB I think I understand it better now.

I've got some questions for you, not necessarily as in I want answers (just in case you think I'm being rude again), but rather to help you think about the strategy:
* What are the average returns of the strategy if you run it only when LYC is >200MA?
* What are the average returns of the strategy if you run it only when LYC is <200MA?
* What are average returns if you only run it when todays Close is N (also test N*2, N*3) ticks (ATRs?) higher (also test lower) than yesterdays Close?
* What's the (average 1, 5, 20 day) correlation of LYC (or whatever stock you choose) to the index? Are returns better or worse if you only trade on high (or low) correlation? Above or below 200MA?
* Are returns better if you use a time based exit of 2 instead of 1? Usually they are.
* Are returns better if you exit using a favourable price (e.g. highest high of last 2 days?) instead of a time based exit? Usually they aren't but sometimes a real edge can be found using favourable price exits.
 
Gringotts Bank said:
skc, it helps greatly to know in advance what sort of returns you can get if you can pick the bottom. It helps to know which stock to trade this way (some it doesn't work on). And it helps to know how close to the Low you have to get to be profitable. All these things are backtestable.

When tested, they tell me:
1. the returns are huge if you can read the depth to within 3 price levels of the low on LYC.
2. the returns are pretty close to zero tending negative if you are 4 or greater levels away - in other words, if a low of 1.16 has been already printed, don't buy at 1.19, put in a buy at 1.175 and hope to get filled. Otherwise leave it for that day.

Futures would be a better instrument, as you say. This thread is about how one might trade a stock intraday.
Click to expand...

Can you tell what is the degree of return vs accuracy?

I suppose if you have 250 trades per year and each correct pick gives you 15 ticks on average. While an incorrect pick yields a loss of (say) 4 ticks (assuming a stop), you can quickly work out what is the "breakeven" accuracy required?
 
skc said:
Can you tell what is the degree of return vs accuracy?

I suppose if you have 250 trades per year and each correct pick gives you 15 ticks on average. While an incorrect pick yields a loss of (say) 4 ticks (assuming a stop), you can quickly work out what is the "breakeven" accuracy required?
Click to expand...

Now you're talking. That's the real crux of it, and the reason for me paper trading it. The answer is no, not until I have traded it for some weeks.

I expect that if 300% pa and miniscule drawdowns is the result of getting all trades within .015, and I get 4 out of 5 right in every week, then I should be able to do ok. This will be determined by two things - 1. my ability to pick the bottom consistently (so far so good), and 2. what sort of size losses occur when I don't pick it. I have a feeling the odds would be in my favour.
 
Mon___B:1.15___S1:1.16___S2:1.15
Tues___B:1.16___S1: 1.17___S2:1.15
Wed___B:1.17___S1: 1.14___S2:1.19
Thurs___B:1.145___S1:1.19___S2:1.17
Fri___B:1.19___ S1:????____S2:1.205

Today I missed the entry, but got lucky with a strong close.

Both exit methods (S1 and S2) profited across the week, which was a pretty dud week for LYC all up.

S1: $2000+ potentially
S2: $700
 
Gringotts Bank said:
Now you're talking. That's the real crux of it, and the reason for me paper trading it. The answer is no, not until I have traded it for some weeks.

I expect that if 300% pa and miniscule drawdowns is the result of getting all trades within .015, and I get 4 out of 5 right in every week, then I should be able to do ok. This will be determined by two things - 1. my ability to pick the bottom consistently (so far so good), and 2. what sort of size losses occur when I don't pick it. I have a feeling the odds would be in my favour.
Click to expand...

Exactly why you need a stop. By having a stop you elminate the size of loss variable.

Since your hypothesis is to pick bottom within 1.5c of LOD, then the most logical action is to have a stop at 2c below entry. If the stop is hit, put that pick in the failed basket.

If you know the average return of a correct pick (which really is just daily close - LOD - 1.5c) and you know the definite loss of an incorrect pick, you can easily plot the curve of pick accuracy vs return.

Surely you have the data to do this now?!
 
skc: Yeh well that's what sinner was saying, and I agreed it made sense. Just that I don't have intraday data for LYC. to test it properly.

If anyone has intraday data going back say 1 month on LYC, please consider emailing it to me and I can do just that.
 
sinner said:
Cool GB, wish you all the best with this, as the range over the last eight trading days in LYC has been quite narrow I am interested to see how you go once it expands again.
Click to expand...

Thanks. Yeh big ranging days will be a good test. I'll get onto answering those other questions when I've tested them.
 
Since your hypothesis is to pick bottom within 1.5c of LOD, then the most logical action is to have a stop at 2c below entry. If the stop is hit, put that pick in the failed basket.

If you know the average return of a correct pick (which really is just daily close - LOD - 1.5c) and you know the definite loss of an incorrect pick, you can easily plot the curve of pick accuracy vs return.
Click to expand...

The thing is though skc, as I said...using stops will vastly alter the performance of this method, almost certainly to prove the hypothesis "LOD+3 (with stops)" to be unprofitable. i.e. You can't pick bottoms without a crystal ball. (who would have guessed?)

My guess is this will be because stops tend to eliminate any ancillary gains that may come from holding till close even if the pick was wrong. i.e. taking a loss versus taking part in the 1 day returns of an up trending stock will generally be the losing strategy. As GB pointed out, if you pick at LOD >3 then negative returns are correlated with how far from LOD you actually picked when it comes time to close (rather than guaranteed loss - unless you enter multiple times per day).

Am I saying this to put GB or his idea down? No, not really. I can't give GB the intraday data he wants, but I can give him my thoughts on what will happen using intraday data based on similar experiments, that's all.
 
update

Mon___B:1.15___S1:1.16___S2:1.15
Tues___B:1.16___S1: 1.17___S2:1.15
Wed___B:1.17___S1: 1.14___S2:1.19
Thurs___B:1.145___S1:1.19___S2:1.17
Fri___B:1.19___ S1:1.21___S2:1.205
Mon___B:1.22___S1:1.21___S2:1.225
Tues___B:1.21___

Paper traded profit: $2345, represents 4.7% return on capital. Doesn't include today, which looks ok at the moment. Calculation based on closing price = sell.

From now on I'll just post the closing price as the sell.
 
Mon___B:1.15___S1:1.16___S2:1.15
Tues___B:1.16___S1: 1.17___S2:1.15
Wed___B:1.17___S1: 1.14___S2:1.19
Thurs___B:1.145___S1:1.19___S2:1.17
Fri___B:1.19___ S1:1.21___S2:1.205
Mon___B:1.22___S1:1.21___S2:1.225
Tues___B:1.21___S:1.22

Paper profit at yesterday's close: $2758

Bid:1.245 <amended twice

B:1.245
 
Ah yes, let's subtract $55 per trade :eek:. [note that my backtests did include commission]

That comes to $1980. Not quite so glamorous now. Still, 4% in just over a week is good if I can keep it going.

The big IF will be when I start putting my own money down. Psychological factors will come into play.
 
Gringotts Bank said:
Ah yes, let's subtract $55 per trade :eek:. [note that my backtests did include commission]

That comes to $1980. Not quite so glamorous now. Still, 4% in just over a week is good if I can keep it going.

The big IF will be when I start putting my own money down. Psychological factors will come into play.
Click to expand...

You need to get a more cost effective broker!

Even if you were with Commsec you would only be paying $19.95 each way. A saving of over $15 per completed trade.

on IB you would save even more
 
investorpaul said:
You need to get a more cost effective broker!

Even if you were with Commsec you would only be paying $19.95 each way. A saving of over $15 per completed trade.

on IB you would save even more
Click to expand...

From memory each position is $50k, which is the brokerage comsec would charge. Makes me think GB is using comsec or something similar. Not the broker to be using when scalping.
 
Top