1.2 Strategy Implementation

The final step is implementing this strategy to create a signal. Let’s begin by reading in the flow files to our R studio workspace.

x.rgn <- mat.read(flow.file.rgn) # GET RGN FLOW PERCENTAGE C:\\EPFR\\daily\\MultiAsset-Rgn-daily.csv
x.FI <- mat.read(flow.file.FI) # GET FI FLOW PERCENTAGE C:\\EPFR\\daily\\MultiAsset-FI-daily.csv

The next step is to ensure both of our flow files have the same rows so that we can combine the two files, like we do below.

x.rgn <- x.rgn[is.element(dimnames(x.rgn)[[1]], dimnames(x.FI)[[1]]), ] # ENSURES RGN MATCHES FI ROWS
x.FI <- x.FI[is.element(dimnames(x.FI)[[1]], dimnames(x.rgn)[[1]]), ] # ENSURES FI MATCHES RGN ROWS

x <- cbind(x.rgn, x.FI) # COMBINES RGN AND FI DATA

One of the first options we have is choosing the universe we want to backtest. EPFR has tested this signal within two different universes of asset classes: Rgn (7 Region Equity) and FI (Fixed Income). For this example, we choose Multi, which includes both of our equity and fixed-income asset classes, but users can also choose to backtest within either the Rgn or FI universes separately.

idx <- "Multi" # Multi/Rgn/FI

1.2.1 Compounding Flows

Next, we set up a variable for our lookback period, which can also be called a flow window. This variable will be the window of time we use to create a trailing compounded daily percentage flow. The lookback period we choose for our demonstrations is 20 days.

lookback <- 20 # FLOW WINDOW (IN WEEKDAYS) - 20 day look back period

Using a function from the library('EPFR.r'), compound.flows() compounds our daily percentage flow over the trailing lookback period for each asset class.

x <- compound.flows(x, lookback, F) # COMPOUND FLOWS

	AsiaXJP	EurXGB	Japan	LatAm	PacXJP	UK	USA	GLOBEM	WESEUR	HYIELD	FLOATS	USTRIN	USTRLT	USTRST	CASH	USMUNI	GLOFIX
20221201	-0.5935744	-0.9011096	0.1759815	-1.231081	-0.3122597	-0.8392634	-0.4846786	0.6630384	0.5921352	1.9514429	-2.125939	0.8615195	7.368753	-0.5797696	1.324865	0.1061790	0.2467439
20221202	-0.4957430	-0.6686647	0.2507892	-1.508537	-0.3909206	-0.7661013	-0.5061060	0.7765973	0.5282304	1.9381506	-2.151642	0.9314150	7.247511	-0.6783618	1.771157	0.1854309	0.2729250
20221205	-0.4947291	-0.7995004	0.0388963	-1.531783	-0.4224801	-0.8483539	-0.5699878	0.9884008	0.4988437	1.6447813	-2.064563	0.2088554	7.170713	-1.3358431	1.654928	0.2551084	0.3154037
20221206	-0.4729045	-0.6938171	-0.0446434	-1.628549	-0.3301458	-0.8645179	-0.6412650	1.0813465	0.5051112	1.4909559	-2.218451	0.4746187	6.925902	-2.1067769	1.780230	0.3172435	0.3053755
20221207	-0.2520256	-0.5759144	-0.1187730	-1.695810	-0.3475549	-0.8860501	-0.6878389	1.0946509	0.5293832	1.3865604	-2.436764	0.3590598	6.852528	-2.4679185	1.562598	0.3119034	0.3425063
20221208	-0.0925901	-0.6546987	0.0569879	-1.612896	-0.4813274	-0.8700669	-0.7834505	1.2062479	0.4497268	0.4233311	-2.837551	0.9271004	6.053509	-1.3176497	2.342189	0.2374214	0.3654252

1.2.2 Total Return Index

We will then import our two total return files and modify them to become one data frame of total index returns indexed so that time moves forward. To do this, we will use the function ret.to.idx() from library('EPFR.r'). We will also use the functions map.rname() to ensure the row names of the matrices line up with our flow file, and ret.idx.gaps.fix() to replace any NA values. Please refer to the library documentation for the complete list of parameters for these functions (tip: ?ret.to.idx(),?ret.idx.gaps.fix() ).

#RGN RETURNS
y <- mat.read(ret.file.rgn) # GET RGN PERCENTAGE RETURNS C:\\EPFR\\returns\\PsuedoReturns-MultiAsset-Rgn-daily.csv
y <- ret.to.idx(map.rname(y, dimnames(x)[[1]])) # CONVERT TO A TOTAL-RETURN INDEX
y <- ret.idx.gaps.fix(y)

#F.I. RETURNS
z <- mat.read(ret.file.FI) # GET FI PERCENTAGE RETURNS C:\\EPFR\\returns\\PsuedoReturns-MultiAsset-FI-daily.csv
z <- ret.to.idx(map.rname(z, dimnames(y)[[1]])) # CONVERT TO A TOTAL-RETURN INDEX   

#COMBINE RETURN FILES
y <- data.frame(y, z)[, dimnames(x)[[2]]] # FINAL TOTAL-RETURN INDEX TABLE
y <- y[max(simplify2array(lapply(y, function(x) {find.data(!is.na(x), T)}))):min(simplify2array(lapply(y, function(x) {find.data(!is.na(x), F)}))), ] # ENSURE ALL PIECES HAVE RETURN OVER THE SAME HORIZON

1.2.3 Subset files

Then, to subset the flow file \(x\) and return file \(y\) to the selected universe \(idx\) of asset classes, the following code is used. During this process, we also ensure that the data structure in \(x\) and \(y\) are completely aligned, having the same column names in the same order.

if (idx == "Rgn") { 
    x <- x[, 1:7]
    y <- y[, 1:7]
} else if (idx == "FI") {   
    x <- x[, 8:dim(x)[2]]
    y <- y[, 8:dim(y)[2]]
}  

* Note: subsetting can also be done when creating the flow and return files

1.2.4 Ranking Asset Classes

Next, we sort each of the asset classes in our universe into five equal bins based on their compounded percentage flow values for the selected holding period. To do this, we will use the function from library('EPFR.r'), called bbk(). This function will output a standardized backtest result.

The bbk() function requires our daily percentage flow data compounded over a desired period and the total return index data. Please refer to the library documentation for the complete list of parameters of this function (tip: ?bbk()).

The first parameter we add is the number of bins we want to use. For our case, we want to use 5 because our strategy is to go long the top fifth and short the bottom fifth.

nBin <- 5 # NUMBER OF BINS

Since EPFR data is published with a T+1 day lag and is released around 5:00 pm EST, we account for a T+2 day delay in our model. Users interested in more timely signals can also use the T+2 open prices for backtesting purposes. Alternatively, EPFR’s Premium Daily offering collects an earlier release of end-of-day data which includes a significant subset of its original fund-level flow information.

delay <- 2 # DELAY IN KNOWING DATA (IN WEEKDAYS) - data takes time to have

It is also important to note that this model will need to be re-balanced weekly. The day of the week the rebalancing occurs is at the user’s discretion. For this example, we will set the day of the week to trade as Friday.

doW <- 5 # DAY OF THE WEEK YOU WILL TRADE ON (5 = FRIDAYS)

Additionally, we also evaluate the returns for different holding periods. The user can input the return horizons that they are interested in here. For this example, we define a return horizon for weekly, fortnightly, monthly, bi-monthly, quarterly, and semi-annual rebalancing.

hz <- c(5, 10, 20, 45, 65, 130) # RETURN HORIZON (IN WEEKDAYS) - holding periods

Now that we have defined all of our inputs, to rank the asset classes into quintiles by their 20-day percentage flow, we call the function bbk() for a one-week holding period.

z <- bbk(x, y, 1, hz[1], nBin, doW, T, 0, delay)

1.2.5 Model

20-day flow percentage ranked into quintiles (computed only where forward returns are available)

z[["bins"]]

	AsiaXJP	EurXGB	Japan	LatAm	PacXJP	UK	USA	GLOBEM	WESEUR	HYIELD	FLOATS	USTRIN	USTRLT	USTRST	CASH	USMUNI	GLOFIX
20221230	2	4	3	4	3	4	5	1	2	5	5	2	1	1	3	4	2
20221223	2	4	3	5	3	4	4	1	1	5	5	2	1	2	3	4	2
20221216	2	4	3	5	3	4	4	1	2	2	5	4	1	5	1	3	2
20221209	3	4	3	5	4	4	4	2	2	1	5	2	1	5	1	3	2
20221202	4	4	3	5	4	5	4	2	2	1	5	1	1	2	2	3	3

Quintile returns over the equal-weight universe

z[["rets"]]

	Q1	Q2	Q3	Q4	Q5	TxB	uRet
20221230	0.1281527	0.0754571	-1.4629180	1.2088600	-0.3776576	0.5058103	1.7369770
20221223	-0.4086635	0.2329289	0.2585288	0.3556384	-0.6346217	0.2259583	-0.5369834
20221216	-1.0962074	-0.4775815	-1.2348837	0.3551931	2.4942756	-3.5904830	-0.1111863
20221209	1.0920910	0.7708809	0.0294642	-1.0666550	-0.7271897	1.8192807	-0.8059708
20221202	0.4451042	0.6385878	0.5724600	-0.5517372	-1.1333650	1.5784691	-0.5930231

Def: TxB represents summary statistics for the long/short portfolio (top - bottom = Q1 - Q5 = overall portfolio returns)

---

1.2.6 Performance

Performance over all holding periods

fcn <- function(retW) {as.matrix(bbk(x, y, 1, retW, nBin, doW, T, 0, delay)$summ)} # DEFINE SUMMARY FUNCTION

sapply(split(hz, hz), fcn, simplify = "array") # WRITE SUMMARIES

	Q1	Q2	Q3	Q4	Q5	TxB	uRet
Weekly
AnnMn	1.5	-1.4	0.9	2.2	-3.5	5.0	3.1
AnnSd	7.9	5.3	6.1	6.1	6.8	12.0	8.6
Sharpe	18.7	-26.4	14.9	36.1	-51.3	41.5	35.9
HitRate	-0.2	-2.9	2.9	3.0	-3.0	1.0	6.7
Beta	-0.6	-0.1	0.2	0.3	0.1	-0.7	1.0
Alpha	3.3	-1.1	0.3	1.3	-3.8	7.1	0.0
DrawDn	-18.7	-35.8	-26.7	-18.9	-55.5	-28.1	-35.6
DDnBeg	20140124	20080104	20100226	20180727	20070525	20140103	20071026
DDnN	312	600	242	91	791	315	71
AnnTo	1185	1821	2567	2008	1395	2580	0
Fortnightly
AnnMn	0.8	-0.5	1.2	1.7	-3.7	4.5	3.0
AnnSd	8.0	5.3	5.5	6.5	6.3	11.8	8.8
Sharpe	10.4	-9.8	22.5	26.5	-58.2	38.4	34.5
HitRate	-1.1	-1.0	3.4	2.3	-5.4	2.1	8.8
Beta	-0.6	-0.1	0.1	0.4	0.1	-0.7	1.0
Alpha	2.6	-0.3	0.9	0.5	-3.9	6.5	0.0
DrawDn	-30.2	-26.1	-17.8	-17.7	-60.1	-32.8	-35.2
DDnBeg	20145512	20075666	20150672	20130374	20070604	20130520	20071064
DDnN	158	305	100	44	394	148	32
AnnTo	911	1260	1632	1322	1031	1942	0
Monthly
AnnMn	0.6	-0.4	1.3	1.1	-2.9	3.5	3.1
AnnSd	8.2	5.0	5.4	6.4	6.4	12.2	9.2
Sharpe	6.9	-7.7	25.6	17.6	-46.2	29.1	33.4
HitRate	-2.0	-2.0	2.8	3.7	-7.1	2.1	10.5
Beta	-0.6	-0.1	0.1	0.4	0.1	-0.7	1.0
Alpha	2.4	0.0	0.9	0.0	-3.3	5.8	0.0
DrawDn	-30.4	-22.6	-16.4	-21.5	-50.7	-29.7	-33.4
DDnBeg	20110790	20115758	20093091	20135358	20070716	20175692	20071040
DDnN	115	78	43	25	187	44	17
AnnTo	676	794	935	795	745	1420	0
Bi-Monthly
AnnMn	0.3	-0.5	0.7	1.2	-1.9	2.3	3.0
AnnSd	7.8	4.9	5.5	6.0	6.4	11.7	9.4
Sharpe	3.7	-11.8	11.3	19.9	-30.6	18.9	31.7
HitRate	-6.1	-3.9	2.0	3.8	-7.3	2.0	12.9
Beta	-0.5	-0.1	0.2	0.3	0.0	-0.6	1.0
Alpha	1.9	-0.2	0.1	0.3	-2.1	4.0	0.0
DrawDn	-33.0	-24.6	-19.1	-19.2	-43.3	-35.9	-30.3
DDnBeg	20106159	20098564	20092837	20091709	20084162	20132922	20072094
DDnN	66	49	24	18	69	41	7
AnnTo	336	371	436	380	380	716	0
Quarterly
AnnMn	0.3	-0.8	0.6	1.2	-1.5	1.8	3.0
AnnSd	8.0	4.8	5.6	5.6	6.2	11.7	9.4
Sharpe	4.5	-16.2	11.2	21.8	-25.1	14.9	32.0
HitRate	-5.3	-3.7	1.0	5.6	-6.5	0.8	15.1
Beta	-0.5	-0.1	0.3	0.3	0.0	-0.5	1.0
Alpha	1.9	-0.5	-0.2	0.3	-1.4	3.2	0.0
DrawDn	-34.6	-26.3	-18.7	-17.0	-36.8	-37.3	-28.1
DDnBeg	20114462	20095990	20086889	20101398	20087622	20134500	20072539
DDnN	40	40	20	11	44	30	4
AnnTo	246	267	301	265	271	518	0
Semi-Annual
AnnMn	-0.1	-1.3	0.8	1.5	-1.0	0.9	2.9
AnnSd	7.6	5.0	5.7	5.7	6.4	11.1	9.9
Sharpe	-1.3	-26.0	13.0	25.4	-16.9	8.0	29.5
HitRate	-2.6	-10.8	4.8	7.3	-7.1	-1.6	16.8
Beta	-0.4	-0.1	0.3	0.3	-0.2	-0.2	1.0
Alpha	1.0	-1.0	0.0	0.7	-0.6	1.5	0.0
DrawDn	-31.4	-31.8	-17.0	-15.8	-31.6	-41.0	-24.8
DDnBeg	20110205	20099885	20100391	20094253	20087249	20117096	20072774
DDnN	18	22	11	6	17	14	2
AnnTo	133	141	152	140	142	276	0

---

Annualized mean one-week returns

bbk(x, y, 1, hz[1], nBin, doW, T, 0, delay)$annSumm # DISPLAY CALENDAR-YEAR RETURNS

	Q1	Q2	Q3	Q4	Q5	TxB	uRet	nPrds
2007	-3.6	9.8	0.9	0.8	-12.6	9.0	5.1	31
2008	29.4	-8.6	7.2	-4.6	-19.0	48.4	-26.4	52
2009	1.5	-1.9	13.7	4.1	-18.1	19.6	24.4	52
2010	1.3	-1.6	-3.5	4.2	-1.3	2.6	8.8	53
2011	-1.3	-1.2	-15.6	9.3	6.2	-7.6	-1.7	52
2012	-0.2	1.4	0.0	1.6	-3.7	3.5	10.0	52
2013	6.0	0.2	-0.6	0.5	-6.3	12.4	4.8	52
2014	-9.5	-1.1	-3.6	5.7	6.9	-16.5	3.5	52
2015	-0.1	2.6	1.3	2.2	-7.6	7.5	-2.7	52
2016	-1.2	-6.5	6.1	1.0	2.3	-3.5	6.6	53
2017	-1.5	-4.8	-0.6	4.9	1.9	-3.4	8.5	52
2018	1.2	-9.7	9.6	-1.8	4.5	-3.3	-5.5	52
2019	-6.8	-3.1	5.8	2.9	1.1	-7.9	11.7	52
2020	20.0	5.3	-11.8	-7.3	-5.5	25.5	4.1	52
2021	-2.6	0.0	0.0	7.7	-7.6	4.9	4.0	53
2022	-8.7	2.3	4.6	0.4	0.4	-9.1	-9.5	52