Misc

This topic contains 62 replies, has 22 voices, and was last updated by  Mark Vincent 1 week, 1 day ago.

Viewing 15 posts - 46 through 60 (of 63 total)
  • Author
    Posts
  • #44670

    Alexander Horn
    Keymaster

    Hi Howard,

    all of our strategies use medium-term trends, e.g. momentum in prices, volatility and cross-correlation within a normally 20-80 days window. Macro-events like the Chinese sell-off in August just make it difficult to detect these underlying trends, e.g. when markets are choppy there is not much of a trend.

    For that purpose we always include hedging instruments like bonds, currencies and commodities in our strategies. In good times they add some performance, in bad times they minimize the impact of the corrections in equities.

    Most important is: Never try to predict or time the market. With the benefit of hindsight of your comment made some months ago: Who would have expected such a strong equity market with so much complacency with volatility being below or around 10 in most of 2017?

    #45176

    mlipsman
    Participant

    The return percentages you show in your tables aren’t what most of us will actually realize unless we sell our entire position each time we rebalance and then buy back the new allocation. Instead, our buy price for each symbol will be the average of all the buys and sells we’ve done so far, which becomes increasingly cumbersome to calculate as time passes and new buys and sells have to be accounted for in calculating our average cost. Is there an easier way to do this, even if it’s a good approximation rather than an exact figure? Within .5% or so would be fine. I’d just like to know if I’m winning or losing each time I rebalance, and by approximately how much.

    #45177

    reuptake
    Participant

    Actually the prices are closing price from the day BEFORE we get signals, so it’s impossible to achieve the same results.

    #45192

    mlipsman
    Participant

    OK, but I’d still like to know if there’s an easier way to calculate it than figuring an average cost based on all previous buys and sells of that symbol.

    #45194

    Alexander Horn
    Keymaster

    Hi mlipsman, You´re normally interested in the return over a certain time period, let´s say since you started investing in a strategy, or since beginning of the year. This is why we publish the strategy returns YTD, 12 months, etc. In the charts you can even set your own time frames and see the return. For a portfolio of strategies you can do the same in the Portfolio Builder.

    These returns always consider the dividend adjusted and fully re-invested (compounded) performance, and the varying allocation to the underlying ETF. Therefore they considers the average price you realized since a certain point in time. From that perspective it is not relevant if you calculate returns as re-balancing only the needed shares, or complete sell and re-buy (excluding transaction costs, which would hit you more if you did the latter).

    So when looking at your portfolio since a certain time, and trying to match your performance to what we report, just take the compounded return over that period, and you see it will match – with a slight +/- due to different execution prices and transaction costs.

    Not sure if I got and answered your point, if not please let me know.

    #45195

    Alexander Horn
    Keymaster

    Hi reuptake,

    think we had this topic before, so same answer. Yes, we use same-day-close prices to calculate our signals and performance – and indeed our signal subscribers can only execute at next open – which indeed leads to small variances.

    But as stated before, according to our backtests this variance in performance is in a range of +/- 0.5% p.a. – in some years even higher than what we report. As we´re advocating for a “lazy” monthly rebalancing within the first or even second trading day, we stick to the current way of reporting, which we feel is transparent.

    If we´d use next-day-open prices for calculating performance, this would be not in line with the prices we use for cutting signals, not allow us to immediately report performance when we send out signals, and make things more complicated.

    Lastly, and in big part to your request, our QuantTrader subscribers are able to cut signals and re-balance whenever they like – even intraday. Personally, being lazy by nature, I don´t do this, and also see no added value in my account.

    #45230

    mlipsman
    Participant

    Hi Alex,

    I’m not sure if you answered it either, so let me try to clarify what I mean.

    Suppose I have $10,000 in MYRS, which is at a 50/50 allocation. ZIV is $50, TMF is $20. I start with 100 shares of ZIV at $50 and 250 shares of TMF at $20. My average cost per share for ZIV is $50.

    Two weeks later, ZIV is at $53, TMF is at $18, and the allocation goes to 60/40.

    I sell 56 shares of TMF at $18 (=$1008) and buy 19 shares of ZIV at $53 (=$1007).

    My average cost per share for ZIV is:
    100 shrs @ $50 = $5000
    + 19 shrs @ $53 = $1007

    = $6007/119 = $50.48

    Your charts would presumably show a symbol return of $(53-50)/50 = .06 = 6%
    However, my return would be $(53-50.48)/50.48 = .05 = 5%

    Two weeks later, ZIV is at $47, TMF is at $23, and the allocation goes to 40/60.

    I sell 21 shares of ZIV at $47 (=$987) and buy 43 shares of TMF at $23 (=$989).

    My average cost per share for ZIV is:
    100 shrs @ $50 = $5000
    + 19 shrs @ $53 = $1007
    – 21 shrs @ $47 = $987

    = $5020/98 = $51.22

    Your charts would presumably show a symbol return of $(47-53)/53 = -0.11 = -11%
    However, my return would be $(47-51.22)/51.22 = -.08 = -8%

    So every time I rebalance, my average cost per share changes: it’s the total dollar amount of shares I’ve bought less the total dollar amount of shares I’ve sold, divided by the number of shares I have left. Your gain/loss is based on the change in price of 1 share. My gain/loss is based on the change in my average cost of 1 share compared to the new price and may not be anywhere near your figure.

    What I’m wondering is if there’s an easier way to calculate (or approximate) my gain/loss without having to figure the average cost per share every two weeks or month, because the calculation gets longer each time.

    Thanks,
    Mark

    #45260

    Alexander Horn
    Keymaster

    Your calculations are correct from what I see, but our way is slightly different. Instead of calculating in avg dollar and share terms, we simply re-calculate an internal index by multiplying the % invested with the symbol return. The outcome is exactly the same, benefit of using avg dollar and share terms is that’s it more intuitive, but indexing is just simpler.

    Attached an Excel sheet with the comparison of both following your example, I ignored full-shares requirement.

    #45262

    mlipsman
    Participant

    Great, thanks! I knew there had to be an easier way.

    #45277

    mlipsman
    Participant

    Actually, looking at your spreadsheet more closely, it appears that the number of shares to own (cols F and G) is determined by the total investment (col J). But the total investment depends on the number of shares (times the price of each), and the number of shares depends on the total investment. So it seems a bit circular.

    It looks like the procedure would be this?
    1. Calculate performance (cols P and Q) based on the change in price (cols D and E).
    2. Multiply performance this period by the investment last period to determine the updated total investment (e.g., J4 = Q4*J3).
    3. Use the updated total investment to determine the new number of shares to own.

    #45289

    Alexander Horn
    Keymaster

    Hi Mark,

    yes, that is the best way to describe. Sounds silly, but performance at the end is what you have in your account divided by what you had in the prior period. The new shares to own is total account value multiplied by new allocation %, divided by share price. Sorry if the sheet made it look more complicated :-)

    #45318

    reuptake
    Participant

    Thanks for the answer.

    Could you maybe recommend what is the easiest way to measure the “real” performance of strategy? I’m using IB, and reports I can generate are bit overwhelming. And I’m under impression (which may be false) that almost all of the time I got significantly worse performance than one that is published on the website. Maybe there is some shortcut? Ready made Excel sheet would be great (in which one would enter buy/sell prices, allocation and get the performance). The impression of underperformance is because most of the time (again: it may be wrong) I got “worse” price than close price.

    And another question: what should I do if I cannot trade during first days of some month? This will happen in November. I’m leaving for a 2 weeks of holidays in Mexico and I really do not want to take my laptop with me. I can trade on October 27th. Should I just get QuantTrader and trade earlier? Is there a possibility to get the signals earlier (QuantTrader is a bit of a hassle for me, as I don’t have PC at home).

    #45585

    mlipsman
    Participant

    Hi Alex,

    Where on the spreadsheet would you account for money added to or deducted from the original investment? For example, in week 6, if you wanted to add another $10,000 or cash out a couple of thousand? It can’t really be added to the Total in column J, because that would skew the performance.

    Thanks,
    Mark

    #45587

    Alexander Horn
    Keymaster

    Hi Mark,

    you just need to account for the add/withdraw in the performance calculation, so add or subtract both in the past and current period. See example below. Hope this is what you’ve been looking for.

    #46111

    mlipsman
    Participant

    Hi Alex,

    I set up your example in Excel, and it seems that it has the same problem of additions/withdrawals skewing the performance. For example, if you delete the 10000 in C4, the performance in E4 goes from 10% to 18%. Maybe I’m wrong, but I’d consider 18% the “real” performance, since the reduction to 10% occurs only if 10000 is added. To me, it seems that performance should be what the market gave us, not influenced by any additional money we invested or withdrew. Is there a way to account for the additions/withdrawals so that they don’t change the performance?

    To give an analogy, if I’m a merchant and I buy a product for $1 and sell it for $2, I make a 100% profit. That’s my “performance.” It doesn’t matter if I invest $10 or $100,000 in a supply of the products to sell–the *amount* of my profit will change, but the percentage/performance remains at 100. Only if I start selling them for $2.50 will my “performance” increase to 150%–again, independent of how much I invest in having stock on hand.

    So my question is, is there a way to separate the amount that we invest from the return that $1 produces from the investment?

    Mark

Viewing 15 posts - 46 through 60 (of 63 total)

You must be logged in to reply to this topic.