Скачать с ютуб How to Build a Backtest в хорошем качестве

How to Build a Backtest

http://EchoPartners.com/videos How to build a community bank asset liability backtest. Hi. I'm Howard Lothrop, host of Echo Partners TV. If you've heard me speak, or seen any of my materials you probably know that I'm a big fan of the back test. In fact, I think the high quality back test might be the most important part of your entire asset liability process. So how do you construct a good high quality backtest? Well first you start with a good high quality forecast. What back testing does is it compares the asset liability models forecast of interest rate risk and changes to the actual interest rate risk and changes in a given period. So assuming we're looking at a quarterly period, let's talk about how that works. You want to have a line item by line item forecast of the yield earnings for all of your assets and the interest cost of all of your liabilities forecasted over this upcoming period. Now, since we're talking about a static back test, this assumes that there is no growth of the bank, that the asset liability mix doesn't change, and there is no management intervention or strategy that changes things appreciably. So what does this mean? What it means is that the forecast reduces almost directly to your net interest earnings and expenses over the past period. So a good starting place, if you don't have a high quality forecast on hand, is to look at your quarterly results from the most recently completed quarter. Then tou take this forecast and you want to add those things that we didn't know. What are three things we didn't know with certainty at the beginning of the forecast? We didn't know how much the bank would grow. You may plan for growth or we might have a budget for growth but the question is how much did the bank actually grow. So calculate that and gross up or gross down, if the bank shrank, all of your asset liability line items by the same amount. Next let's look at how your asset liability mix changed. You know how it is. Some some quarters you might grow your HELOC portfolio more. Other quarters, it might be your consumer loans or your CRE. So some of these assets and liabilities will grow while others will shrink. We didn't know that at the beginning of the quarter so adjust each and every line item by the amount that line item grew or shrank in the balance sheet over the quarter. Next let's look at how rates actually changed. Again we didn't know that at the beginning of the period. Also, this is much like the asset liability mix in that the yields or interest cost on any given asset or liability line item might be going one way or a another or perhaps growing or shrinking at a different rate for each item. So here again, some might be adjusted upward while some might be adjusted at a faster or slower pace. You must do this on a line item by line item basis for each asset and liability. So one you compltete adjusting for these three items that you couldn't know, you have what we would refer to as an adjusted forecast. Compare this adjusted forecast with your actual interest income, interest expense and net interest income again on a line item by line item basis for the quarter you just completed. You see the adjusted forecast, once you have evaluated the change in an assets and mix and in rate, becomes the forecast that your model would have created had we known with certainty all the things that were unknown. So when you compare this adjusted forecast with the actual what you're left with is is what accountants would call your unexplained variance. It's a measure of the error in your model. Compare this on both a percentage basis and a dollar basis, the amount of this error, on line item by line item basis. Now you have to use both percentage and dollar basis because it if you just look at percentage some very small dollar item might to go from 0 to 1 you have an outlandish percentage change. Or you might have to look at dollars because you might have a small percentage change that maybe, if it's in your loan portfolio, it makes a huge dollar impact on your earnings. Generally speaking at the end of the day if your interest income, interest expense and net interest income numbers on your adjusted forecast are within 10% of your actual numbers this would qualify as a high quality back test. Now that you know how to build it, go ahead. If you need a little help on that, or you just want to see a sample, let me know and I'll send one over to you. Thanks a lot and we'll see you on the next episode.