Advisory Services for proprietary performance tracking and grouping in US equity
Donnerstag, 18. August 2011
How does cluster analysis work with equity
My daily forecast has 3 components
1) the simplest part is the technical -analysis running over daily quotes ohlc
time-series 150 day, indicators trigger automatic every day
using:(pasr,bolinger,ema,sma,rsi,macd,st)
2) part two is statistical sampler for (today,2day,5day,10day,30day,all-time)
3) k-means cluster algorithm which test all symbols as the technical analysis,
to split all companies into 10 cluster and after selecting only those symbols
within the top cluster.
k-means uses a matrix to create clusters
like this [50dayCHG%, 200dayCHG%, %todayCHG%, DIV, M.CAP]
for each symbol
so to make them appear in the list
they have be in top cluster + technical signal
and the statistics must show earlier trades making money
for that combination
:)
1) the simplest part is the technical -analysis running over daily quotes ohlc
time-series 150 day, indicators trigger automatic every day
using:(pasr,bolinger,ema,sma,rsi,macd,st)
2) part two is statistical sampler for (today,2day,5day,10day,30day,all-time)
3) k-means cluster algorithm which test all symbols as the technical analysis,
to split all companies into 10 cluster and after selecting only those symbols
within the top cluster.
k-means uses a matrix to create clusters
like this [50dayCHG%, 200dayCHG%, %todayCHG%, DIV, M.CAP]
for each symbol
so to make them appear in the list
they have be in top cluster + technical signal
and the statistics must show earlier trades making money
for that combination
:)
Abonnieren
Posts (Atom)