Scan to find stocks hitting a support level

srm

hello folks, is there a clever way of writing a scan that finds stocks at a support level. for exp - Intel (INTC) hit the 18.61$ support level 6 times since 8/2024. it would be nice to catch such support level repeated visits. thx!

markd

If you know the price level, it's easy. So, for instance, if the level is 18.61, write a scan to catch price going below, say, 17.

[symbol is 'INTC']

and [17 x low]

This will only give results on the day of the crossover.

If you do

[symbol is 'INTC']

and [low < 17 ]

you will get INTC on every day that condition is met.

If you don't know the support level, you have to write something more generic. Stochastics tracks where price is in its range over some period of time. When it gets below 20, it's near the lowest price in the range, which may be support, especially if the trend is up. But it can also work when price is in a range.

In the case of INTC, it's in a range along its 52 week low, with price rising and dipping back to near that level.

To see if a stock is near its 52 week low you could write, with dips near that level,

[group is sp500] // or your list, or another universe

and [Fast Stoch %K(251,1) < 20] // 52 week (251 day) range is in bottom 20 %

and [Fast Stoch %K(20, 1) < 20] // 20 day range is also in bottom 20%

but, this doesn't distinguish between falling stocks ( a lot of those right now) and ranging stocks, so it may help to add

and [Lower Price Chan(251) = 100 days ago Lower Price Chan(251)] // play with the 100 parameter

Another possibility for finding support, but not necessarily at the same level, as the stock rises off its 52 week lows (possibly starting an up trend) is

[group is sp500] // or your list, or another universe

and [Fast Stoch %K(251,1) < 20] // 52 week (251 day) range is in bottom 20 %

and [Fast Stoch %K(20, 1) < Fast Stoch %K(251)] // Shorter period % K below longer %K

Of course, none of these are perfect, and as usual with pattern-seeking scans, you will get a lot of junk and, depending on the market, more or fewer promising hints. Right now, with the market falling you will get fewer hits. To test out these scans, run them from back dates (use the calendar drop down at the top of the scan window to select random past dates). To filter bad entries, you could try not entering until price closes above the lowest high of the down leg, and then enter on a lower low in the 'expected to continue' up leg with a stop below the lowest low of the down leg. That would keep you out of most entries that keep falling.

srm

Thanx for your input! will teak the stock approach. also seeing if chatGPT can help! (it writes good stock charts scans!)

markd

Thanks for the chatgpt tip. I didn't know that.