How would you test for trendiness?

Quote from jerryz:

what if shorter term volatility acclerated because the price is falling in that time frame, but longer term time frame says prices are rising? in this case the measure would say the trend is acclerating, but which way?

if yes, what measure did you use?
More...

=====================
Yes /up & to re- phrase Garghur2 bull market ''you know'' paragraph;
you are describing a classic pullback to support area, in an uptrend bull market.

Now thats not a prediction , but look @1year SPY chart day of OCT 4-5; looks like a normal correction[pullback], most all of them above 50 day moving average.

Did go further past 50dma than usual, OCT 4/ following;
but obvious up trend still in force.

Its not a prediction cause one of these months/years bear trends happen.
:cool:
 
Quote from jerryz:

thanks. this is interesting, same kind of idea as the sharpe ratio. the higher the number, the smoother the trend.

however, say we have a market that moves like this:

moves up 20 days in a row, then
moves down 20 days in a row, then
moves up 20 days in a row, then
moves down 20 days in a row.

the ending price is now the same as the starting price. (assuming same increment each day)

in this case, both of the ratios would be 0. but from a trading perspective, this market is great. 4 nicely predicable trends. the two ratios won't capture this.

are there statistical measures that would measure this?
More...

No.

Edit: There are, however, several excellent methods to keep you on the right side of the market. Elite has documented them all.
 
As an answer to the original question in this thread, I would suggest looking into the Hurst exponent. Trendiness would be indicated by values significantly in excess of 0.5.
You can measure it using the "rescaled range" calculation.
This is very much a global, statistical measurement. Note that it can change over time for a particular market.

I have trouble linking this calculation directly to trading system performance. In general it's better to start from precisely what you want, so rather than say "I need a trendy market", I think it's better to say "I need a market that has this kind of distribution of price changes" or something like that.

Good luck with your research :)
 
Quote from jerryz:

thanks. this is interesting, same kind of idea as the sharpe ratio. the higher the number, the smoother the trend.

however, say we have a market that moves like this:

moves up 20 days in a row, then
moves down 20 days in a row, then
moves up 20 days in a row, then
moves down 20 days in a row.

the ending price is now the same as the starting price. (assuming same increment each day)

in this case, both of the ratios would be 0. but from a trading perspective, this market is great. 4 nicely predicable trends. the two ratios won't capture this.

are there statistical measures that would measure this?
More...

The SMA(200) of ADX(14) series would measure this. Markets with a higher value would be more "trendy" (over the last year) in this respect.
 
Although jerryz didn't ask it specifically, I guess what he'd like to know is how to identify a stock (or other asset) that has a history of trending nicely, so he could trade it with a trend following system. Most of the answers he got tried to identify if currently that stock is in a trend, and this result will lag the price action at the point where it might be unusable.

I suggest using an indicator like Mesa Sine Wave (developed by John Elhers), or ADX, and look back over (maybe) ten times the time frame you're trading.
Quote from Steveyd:

The SMA(200) of ADX(14) series would measure this. Markets with a higher value would be more "trendy" (over the last year) in this respect.
More...
 
You must log in or register to reply here.
Back
Top