|
PureBytes Links
Trading Reference Links
|
Well, Mark, I 've been around so long I may be forgetting some things.
I'm not sure what you mean, but I may be doing it. I use unsplit
EOD data, and have adjusted all my indicators to transit a
split without a large discontinuity by multiplying the prices
after the split until the period ("length") of the indicator.
"Length" bars after the split all prices used by the indicator
are continuous (normal?), so no post-split multiplication is needed.
This ain't perfect, but it's close.
donc
>All you need to do is Forward Normalize the Price input of all your
>indicators this will solve any problems with splits and gaps. I
>figured you knew this by now, you've been around a long time. mb
>
>
>
>:I think momentum sucks. Another thing wrong with it is that if you
>:use it for backtesting stocks your results might be meaningless. If
>:a stock is split the historical value changes after a split - what
>:was ($30 - $20) becomes ($15 - $10). Better to convert to a
>:Percent (rate of change) rather than use an absolute dollar value.
>:Even converting to a slope, as Bob suggests, is inaccurate. The
>:previous
>:$10 difference, for example, per month, becomes $5 per month.
>:Not conducive to good system development.
>:
>:donc
|