Frederick: We are in another circular reference discussion here. I gave you pretty good instructions the last time we had this discussion on how to program your own model, which requires knowing how to program and set up an Excel spreadsheet. My model is done using the same Excel spreadsheet but I can’t lock the formulas. This happens to me about once a day; I enter something on a box that has a formula in it and out goes the formula, which means I have to re-enter the formula but I have enough experience to know when something doesn’t look correct. If I sent somebody else a copy and they make one booboo they will not know it and the entire algorithm will be screwed up. Then there is the question of knowing some simple but basic statistics and some new appraisal theory. This stuff is not complicated but it requires graphics to illustrate it.
In summary, you need to understand the underlying appraisal theory which requires a simple understanding of trend lines and least sum of the squares regression, you need to know the formulas so you can re-enter them and be able to spot when something goes wrong. If anybody knows enough to do the above, then I have given you in this post all you need to know except the theory to make your own. If the formula cells could be locked so they can’t be inadvertently erased we could test it out on some people. I use to have a spreadsheet program that you could do that. The proper forum for learning this would be in a classroom but I am not inclined to get into the CE business, write a book, or sell software, because I don’t want to and there is not enough market for it anyway. I love being a small time country appraiser that stays at home and plays with the grand boy. I don't have anything to sell, I have shared my experience for free, and I don't feel that I owe anybody anything. It works for me and that is good enough for me.
There is interesting story about showing my model: This is the only person I have ever shown it to, mainly because nobody locally has any interest in seeing it. A few years ago I was doing a commercial site valuation in NC, just a few miles from VA., and I was working with an appraiser that lived in that NC county. He had a degree in real estate and he had worked for an MAI for about 5 years. He came by my office to drop off some sale information and I gave him a demo of how to appraise SFR using regression analysis. When he left my office he seemed to be in a state of shock. He said: “You have really shown me something today.” For the next six months I didn’t see or hear from this guy. Then one day another appraiser came by my office and I asked him what happen to Thomas because I hadn’t seen him for a while.” He smiled and said: “He told me he saw your regression model, said to himself there is no way I can compete with that, gave up appraising, and is now selling AFLACK Insurance.”
Blue 1: You are absolutely correct about using regression to prove what you are doing. Believe it or not, I have not used my regression algorithm in about a year for the simple reason that I don’t need to use it any more. I used it for about 5 years experimenting and now I know which factors influence price and which ones don’t so I can use a much simpler model that resembles the existing marketing grid. That is another reason I am reluctant to give somebody a copy without instruction because you have to know which factors are most important and in which sequence to adjust and if you have not experimented with regression you probably don’t have a feel for it. You have to know when to stop adjusting too. In my view the future of this method is in appraisal reviewing which is an even more limited market. It will have to evolve into existence for reasons Mike gave above: “Old dogs are reluctant to learn new tricks.”
