Discussion in 'Automated Trading' started by HotTip, May 11, 2010.
Sorry. I assumed that scalping by definition was a reversion strategy.
Indicators are definitely:
* ATR - if tthe ATR goes significantly crazy, things are bad.
* Tick jumps. Depending on instrument (Watch futures, please), if things do not move nicely, the market is definitly not normal. YM futures had jumps of 150 dow points as fas as I could see.
* Bid/Ask total volume over like 10 levels. In the YM bids and asks basically disappeared, plus the spread.
The last is a good indicator when a market gets dangerous - if bid/ask volume is really low, a stop loss cascade can get REALLY ugly, and that can be a normally non-move.
Together they could definitely form a breaker circuit.
Do you not find the cost of "quit before you reach your loss limit" when spread tells you market is different, is prohibitive.
The cost of exiting a trade, and missing potential opportunities, is usually higher than the benefit of getting out before a bad move.
Do you find otherwise in your backtesting?
I think regardless of the attempt to tell when things may be different, sudden shocking news may not be detectable
For example, if there is a story breaking that some kind of nuclear exchange occurred (even false), then the serious professionals may be way ahead of you.
You could wind up bankrupt before you had a chance to sleuth
There really is no excuse for people who think minimal trading margins are a reason to be leveraged 50:1 on s&p's. There are always going to be black swans. The best protection is either to keep leverage to a reasonable level, buy options instead of instruments, be very diversified/currency neutral, etc.
The meltdown of Wall Street shows that even the smartest still can train wreck,
no, you can always set place new orders to get back in when the market calms down. this is a scalping thread...how much is the cost? a tick
it depends on your strategy, i assume you dont care if the market makes large moves as long as the liquidity is intact - which means bid/ask spread. In that case what you said would work, something like below.
I think when shit hits the fan, it's best to use the most simple monitors as oppose to any of those fancy indicator, and you cant get any simpler than raw tick data as they are streamed in. I wouldnt use/trust anything else, and if tick data has gone bad, then you know it's over.
if ( abs( tick1.bid - tick1.ask) > (avg+constant) AND abs( tick2.bid - tick2.ask) > (avg+constant) etc.. ) OR current_tick.quote_wait_time > x time_constant )
send bing alert
business as usual
you can have multiple feeds hooked up, using worker threads to monitor each of the feeds independently, if most of them start to turn red, stop trading. This assume your system is listening to tick data from the vendors.
Being out by 1 tick in FX isn't guaranteed...
I guess markets are different.
Sorry - another thought.
If you are picking up that market is "different" surely you are less likely to get out for 1 tick?
Would spreads not be likely to have widened?
The DOW was dropping 100 points a minute. There was no question that we were going down.
I think it comes down to what do you want to do while the market is crashing.
My traders did exactly what they were supposed to do. The short traders rode the market down and closed all short positions profitably at or near the bottom. The long (swingtrade) traders rode it down and added to their positions at or near the bottom and then rode the market back up.
So I guess the answer to your question (for me anyway) is to be prepared ahead of time for what can go wrong, so that you can take advantage of it.
Some people say last Thursday as the worse day ever, while I thought it was vintage; the best stock day I've ever seen. I would be surprised if we ever get to see that happen again. That day was made for robotic traders. If you manually trade and took a break for 5 minutes, then you missed it, but the robotic traders were all over it.
Separate names with a comma.