[pcabc]

Market breadth overview

This looked somnewhat negative during the week but seems to have got stronger at the weekend.

Major indices


US Sectors


UK Sectors


Commodities


Miscellaneous

I am reading your analysis every week, but i am wondering how you calculate / determine your signals (buy, alert, ...)?

This is fairly complex and they are only an indication to myself whether it might be OK to buy, hold or should consider selling.  I decided to share this.   Don't consider it any more than something some random person on the internet has produced.  It is farily ad hoc, but the components make some sense.

I lost a good feed for breadth data a while back, but I have a data feed for end of day data.  From this feed I group stocks into areas, e.g. Russell 3000, UK sectors etc based on publically available free data sources.  However, since this is manual it is somewhat infrequently updated. 

I store the data in a database and overnight, for each group of stocks I take breadth measurements for I have a script run and calculate the number of new highs, new lows, advancing / declining positions etc.  This then goes back into the database.  This data is then used to calculate the more familiar indicators, the Advance / Decline Line, New Highs minus New Lows etc. 

I have four time periods, very short term, short term, medium term and long term.  I then compare these breadth indicators against their moving averages, whether they are above or below key levels etc.  So a bullish indication is +1 and a bearish one is -1.  I consider which time period each indication is within.  So, using price as an example, the MACD histogram being above the zero line, meansing the MACD line is above its signal line, is a short term +1 score, below is -1.  The MACD histrogram being in an upward slope is a much more rapid indicator so it is a very short term score.  Anything based on an MA or 100 say, or price being above its 50 day SMA is a medium term score.  Anything relating to the 150 and 200 day SMAs is a long term score item.  But I apply this to the A/D line, New Highs minus New Lows, % stocks above their EMAs etc.

What this gives me is something akin to Weinsteins weight of evidence, but in four time periods.  I did this as my database does no easily give me weekly charts so I get differing time periods for breadth in my consideration.

My overal scores are fairly ad-hoc, but obviously, if scores in all time periods are bullish then the overall result is bullish or 'buy'.  I note that I am looking more at whether overal breadth scores are increasing in the medium and long term scoing contributions as a strenghtening in market breadth is what I am looking at.  The short and very short scoring periods have this built in.

A word of warning.  My performance has best been described as patchy at best.  Several things seem to be causing this.  At one point I was only buying when breadth (not split into these time periods) was strongest leading to me only buying at tops, which does not work, the markets turned and i was left with losers.  Further I think my costs were too high grinding accounts down and some accounts just had too much lag or lacked useful order types.  So I am doing this as learning and I am still learning.

I'm finding the scheme I have come up with keeps me from buying most of the time.  Which generally seems to be a good thing.  I'm only buying now when may be a lot of market strength, in all timescales (when I behave and follow my own breadth indicator).  Sometimes this works, but the market can still drop back, so it is not fool proof.

So overall this is somewhat ad-hoc, but hopefully built on sensible foundations.  If you are interested and have suitable data source that suit your own way of working then I'd suggest that you re-read the breadth chapter in Weinstein's book and have a play around and see if you can get a useful indication yourself.