(I suggest you read the post below and then read the update here.)
Well, almost never.
While the late-apex cornering technique is a staple of road-racing and track-day driving, it has almost zero applicability to autocross. Why? Autocross almost never has an acceleration zone long enough to make up for what you give up in the corner for the late apex. Occasionally, yes, you will want to late apex, but not very often. Autocross is mostly made up of connected curves of varying radius. Usually it’s best to simply take the shortest path from one to the next, as Piero Taruffi stated in the first-ever scientific book on race-car driving in 1959. He was right then, he’s still right.
Late-apexing is done on track for various good reasons, but the only one related to saving time is to increase exit speed off the corner by “lengthening” the straight. The increased exit speed is carried down the ensuing straight whose average speed is now increased, reducing lap time. This is the only occasion to use a late-apex: when the length of the ensuing straight is long enough to save more time than lost in the corner.
Be careful not to confuse late-apexing with the technique of “back-siding” the cones. Back-siding a cone is not late-apexing. Back-siding a cone is a result of the racer deciding where the beginning and end of the corner are and on what radius. When he tells you he plans to backside a particular cone, it means he has decided that that cone is on the minimum radius but not at the apex of the “corner” he has imagined out there among the orange cones. When he passes it he plans to try to run over the base of the cone with his rear tire, meaning he is wrapping around it in order to be going in the best direction toward the next feature. By definition, therefore, the backsided cone is not an apex in the road-racing sense because the car continues to turn hard well past it.
Now, I suppose one can be perverse and “decide where the corner is” and decide to late apex it and decide that the cone marks the late apex or is at least within shouting distance of it. In that case I admit that you have sort of backsided the cone and pulled off a late apex simultaneously. Good luck with that. Please go back and read the title of this post again.
Be aware as well that designating an offset cone as your corner exit and deciding you want to be accelerating at that point toward the next offset cone is also not late-apexing. (I plan a later post on that subject, complete with diagrams and spreadsheet calculations. I know: You. Just. Can’t. Wait.) How you performed that corner, what path you took, etc. determines whether you late-apexed it or not, not what you were doing as you passed the cone marking the exit. Did you take an extra-long, small-radius, time consuming path that allowed you to increase the length of the straight you created toward the next cone, rather than get to the cone on the shortest, fastest geometric radius? Then you did, indeed, late-apex that baby. You were probably wrong. And slow. 2019 Update: Adam Brouillard in The Perfect Corner has shown that the car that takes the “extra-long, time consuming path” never catches up to the car that didn’t, no matter how long the straight that follows.
Remember that in good autocross course design, the exact location of the corner and even it’s radius is to a large extent at the discretion of the racer to decide. Yes, most of us have seen courses completely lined on both sides with a zillion cones, all marking a path 12 to 20 feet wide. Unfortunately, some organizations still do courses like that, but the top levels of the sport have moved beyond such drudgery. People who always race on such courses will never get FAST at high-level autocross. If they should attend an event where the path is not dictated to them, or even if it has only one or two sections not dictated to them, they become lost, dazed and confused. It’s not the heat, it’s the course.
This is one reason why autocross is so challenging and so rewarding. Every course is different and full of “corners” that the racer has to look ahead and “find” before he can even analyze and then drive them! This may also be why road-racers typically have a hard time adapting to autocross if they didn’t start with autocross. On a track the “corner” is more of a fixed, known quantity. There is great skill in figuring out how to take it at ever greater speeds, how to pass someone in it, and how to not get passed in it. Plus, the same corner is different in different conditions. However, the skill of “deciding where the corner is and what it looks like” doesn’t get developed.
Saving Time - An Autocrosser's Blog 
Ed, Thanks for doing this blog! Maybe I won’t have so many questions next time! Of course, you might have just generated a bunch of new ones!
Btw, I really like the clean layout. GlennB
LikeLike
I was under the impression that late apexing was done at the end of a fast section to delay braking as late as possible as well as for turns with decreasing radius For a turn leading onto a straight from a slow turn, don’t you use an early apex to get on the gas sooner?
See this video, for example: https://www.youtube.com/watch?v=kF0aRhD8_uI
LikeLike
I’ve seen that series of videos. Some good info but this example is a red-herring and almost comical. For what is termed a neutral turn in the video we are shown a comparison between wrong and wronger. It is true that the late-apex method shown is definitely faster than the totally messed-up early apex. It does allow a later braking point than an early-apex, but this is not key. If they had done a “geometric” apex, the time would have been faster than the late apex. That is, wide entry, make the apex at the very center point and wide exit. The time from entry to exit will be significantly faster, as countless people have shown.
He doesn’t mention the “real” reason for late-apexing most corners on a road course, which is increasing the speed down the following straight. At least he (apparently) knows that that doesn’t make sense for autocross. (Because that following straight is almost never long enough.) But, he is still wedding to the late-apex being best, which is just plain wrong. This is a mistake that many, many books on autocross have made. This basic mistake is not properly translating the “road-race” formulation, which uses a certain width track, to the more complex autocross situation, where the “track” width changes radically.
It is also slightly deceitful (in effect, not intent) to show an improperly executed early apex as the “bad” way. Yes, what is shown is bad! But not simply because it is an early apex. Done right, you would early apex and slow enough to follow the inside curve around at the minimum radius, then leave the minimum radius and accelerate on an increasing radius, tracking out to the edge of the track. If that early apex is done with a delayed braking + sliding + trail braking, it would have been faster than the late-apex as well. The entry would be quicker and the distance covered quite a bit shorter.
LikeLike
That makes sense.
What I think of as a late apex corner in an autocross is a 180 (or more) with a wide entry and a narrow exit. If you try for the minimum radius by following the inside cones, you practically have to stop to keep from running into the outside exit cones. Sort of like this:https://dl.dropboxusercontent.com/u/91578766/ETR%201%202014/Finish.png The gray line from GPS readings from my data logger. the crude red line is about what I saw a lot of people doing who were slower than I that day. The tight path might actually be faster from corner entrance to exit, but your exit speed is so much slower, you lose all that time and more in the next segment.
LikeLike
Very good example. I think you drove it right, but what you did I do not think was a late apex. The cone you have circled is the “sucker” cone and quite a common design. It sucks the unwary into misjudging the actual shape of this decreasing radius section and losing all momentum when they get to the end of the wall, as you point out. You did it right by taking a path that allowed maintaining speed all the thru to the finish.
LikeLike
With your permission, I’d like to use the picture you linked to as the subject of another blog? I want to draw on it a bit, determine what was or was not a late apex, what that would look like, and also make a suggestion for how you could have been faster.
LikeLike
Sure. And yes, I though of it as a sucker cone too. I hope you clean it up a bit, though. I can’t draw very well and, if anything, I’m worse on a computer. I also think that what you call a late apex isn’t what I think of as a late apex.
I had a thought about the process. You could work backwards. I think you want to backside the inside cone at the exit. Then you have to work out the path that puts your car in that position with good exit speed.
LikeLike