Island Reversals


Established member
Both these sites give a full explanation of island reversals.

More important tho is "do they work?"

I have a simple routine written into AIQ which identifies islands, and then I can back test whether trading them would be profitable. They tend to be more reliable if they occur after a selling climax, or close to support or resistance, or close to a fibonacci level.

I have "pulled" the table that some of you may have read earlier - I discovered some inconsistencies in the results and don't want to mislead anyone with over optimistic results. Will post it again when I have checked it out and i have a table which stand up to close scrutiny.
Last edited:
The crucial thing to note in your results are the ratio of winners to losers.This leads to a few things:
1: You can almost forget about the losers- stupid in reality, but a fact borne by these results.I just know you wouldn't do that, and not to the extent of taking a 10% loss anyway. Each individual trade will have it's own profit exit point as well as stop loss, calculated before entry.
2: Managing your trades on a more realistic daytrading basis, your results will be significantly better that the mechanical system tested here.
3: 4 years of backtesting means that all possible( practically) market conditions have been in force, adding weight to the validity of the mechanical backtest.

I have attached an example of a Top Island Gap. In this case, on could have picked up the gap up , traded to the top, and shorted the reversal, making a packet.


  • dfsgap.gif
    8 KB · Views: 736
Roger, can you email me the EDS, I can do some further work on it and publish the routine and results......
Just been checking the charts for the competition, and I'm going to pick Ryan Air for a short- Island reversal. It's a gamble, because the reversal hasn't happened - yet- but it looks entirely feasable. Historically, it gaps and reverses a lot.Realistically, an entry at 760/755, stop at 770, and out at 705, if reached.Or half out at 705, the rest at 650, if reached. Given their current strength, I suspect that if it goes, it will only pull back to 705 minimum, and this is one of those cases that a trailing stop protecting 90% of profits >5% would be in order.
The Ryanair c'stick shows indecision as the price approaches long term resistance. Check out easyjet in the same sector - also hitting longer term resistance ans showing clearer reversal potential on the daily bars.
mmmm it really was a pick just for the sake of it being an island reversal.. EZJ I agree, BUT the one thing that puts me off is the big buy that went through a few days ago....along with the instructions to the mm's to keep the price up :)
Sorry to be a pain but if this was the Nasdaq i would know exactly what was going on because i could see what was happening on my level 2 screen.
The point here NAZ, is it's a competition enrty for the week...L2 is irrelevant because we are looking for an overall drop for the week- a penny here or there doesnt matter.
As per my post last night, I shorted EZJ at 410 and covered at 395 cfd for a clean 15 points profit. I considered closing only half the short, but for EZJ to start rising again after this pms crash suggests buying strength despite the daily chart. It certainly looked that way on level 2.
I believe you've got to be flexible - it may go down more but I'm happy. Sometimes just as a scalp turns into a swing when it continues to go your way, a swing trade turns into a scalp when it starts to reverse direction. In this type of market you must take what the market offers you, especially when you are trading against the overall market, sector and share direction.
Well done. the gain was twice the percentage of RYA loss.
Yes I would have expected much more than that , given the news.
Anything holding up today should do well short term.
Thanks. Today was interesting as they rose 23p. The late strength yesterday did indeed turn out to be real buying so the decision to cover and take profits proved correct. Sometimes it isn't, but as long as the decision is based on either evidence or an experienced judgement rather than emotion, then the odds are more in your favour. This, after all, is a multi-factorial probability game and all we can do is try and read what's happening - and go with that rather than what we think ought to happen.
I've rechecked my backtesting routine for island bottoms, and the result for looking at islands in isolation is disappointing. The island has to be a 4 week low, but even so, this is not a winning strategy. I use islands myself sometimes, but like to see bottoms near support, and tops near resistance, and if it's close to a fib level, so much the better. A volume spike for a bottom also helps - they seem to work best when they are formed by a selling climax.

Anyway, here's the result of the basic scan. There are plenty beter routines!



  • islands141101.gif
    17.8 KB · Views: 443
Flea - the scan was run on the AIQ UK Groupwhich includes all UK shares which have a record of reliable volume. In practice it approximates to the constituents of the all-share, about 800 shares.

The scan required that the price had to have dropped to a 20 day low, but of course that may only be a 0.5% fall. I need to rerun it with some changes, for instance, requiring that price needs to have dropped by 25% from the highest that it has been in the previous 20 days, and perhaps looking for a volume spike as well. I am surprised at how poor the results are from this simple mechanical test.
I wonder if liquidity is also a problem? Some stocks tend to gap on a regular basis, raising the possibility of islands being formed that are of less significance. I find that gaps and islands are really of any significance where these events are rare.

I wonder if your back test is including some stocks that gap regularly and therefore distorting the results with weak candidates

Scans are usefull for finding gaps and islands but the final analysis still really needs to be done by eye.


Steve Ballard
Yes , illiquid stocks frequently gap all over the place and should be discounted.
It's always worth checking whether the island reversal is at support or resistance - it is much more valid if it is. See my earlier comments on ESZ as an example.
Rather like candlesticks and many other indicators/techniques, the probability of success is improved if another, UNRELATED indicator confirms.
Nice material, Roger - you're delving deeply in delightful Devon :)
Guys - I completely agree with what you all say. The scan was written purely to enable me to identify islands from any day's data, and particularly at what I thought might be market bottoms. It was never intended as a universal trading strategy.

What I tend to do is when the market looks like it is close to support or resistance is to run my island scans (I have one for tops as well) and then to "eyeball" them to look for other reasons to act, along lines already discussed - particularly islands close to support or resistance.

The main reason for posting the result of a trading backtest was to demonstrate that buying an island without any other criteria being considered is not a winner. I did in fact also add a criteria requiring the island to complete below a 90% band, and also for volume on the reversal day to be 50% above the 10 day ma of volume. To my surprise this did not improve the result significantly.

However, islands still form part of my overall strategy, but I did not appreciate until I ran the back test just how much I also rely on supporting criteria.

Thank you for sharing a super piece of research, Roger.
I seem to remember you doing a similar investigation with head and shoulders a few years back. Do you still use that work? Perhaps with H&S bottoms rather than tops these days? :)