This will not work because the rate of binding is affected by both concentration and the binding rate constant.  The latter is not necessarily the same for all antibodies.

I agree with Dr. Ullman. This is not possible in theory as the kinetics of the different antibodies will vary, and even if it was possible (if, for example, you were looking at different concentrations of the same antibody) would be very difficult to do in practice. The rate of reaction is dependent on how the solution is mixed (i.e. *exactly* how you pipette the diluted sample into the plate, how you then shake it etc..) as well as temperature. 

You can estimate the titre with a quantitative assay and a more sparse dilution curve, i.e. rather than doing 2-fold dilutions are looking for the one that shows no signal, you do 10-fold dilutions, look for the one that shows little signal and the one that shows no signal, and interpolate to the level at which the signal drops below detectable level. This saves some sample, but a) is not as accurate and b) still requires you do serial dilutions. 

I do not understand however why you say "I can't do standard curves." . If you are doing serial dilution of a sample, then the dilution at which you lose signal tells you the titre. If you compare that to another sample, diluted in the same way and tested in the same assay, that tells you which sample is more potent. There is not a 'standard curve' involved, unless you mean that you have no reference sample with which to compare your test sample. And if you have no reference sample, how do you know your test titre is meaningful? 

Antibody titers without a standard curve or a reference calibrator is a circular problem.

 Furthermore, the calibrator needs to have the same characteristics (monoclonal vs. polyclonal), K On/Off rate etc...

What is the titer of an antibody ? It is not the quantity (?) but also the avidity and affinity of the antibody as is indicated by Edwin . These factors are method dependant.  It is also important to consider the 'matrix'  if you dilute the sample the same environment in the dilutions is necessary to assure full reaction with the antigen.

@Edwin : I agree that the the rate of binding is affected by both concentration and the binding rate constant. However, binding rate constant should be the same since I use the same secondary antibody to test all 10 sera. The way I see it: ELISA plate is coated with my antigen of interest. I apply serum that contains antibody against multiple epitopes, hence different affinity/avidity. However, the secondary antibody (coupled with AP or HRP) recognize the same constant region of the heavy chain of these antibodies hence the binding rate is the same. Since they're also coupled to the same enzyme, I can infer that substrate consumption is directly proportional to the concentration.

@William: The solution is mixed the same way for every well using multi channel pipettes in the same manner and incubated in the same conditions, so I don't think that the methodology will significantly influence the outcome. And I precisely want to avoid serial dilutions since I need many more ELISA plates to test a large number of sera.

@Maan: antibodies in my sera are not purified nor selected in any manner, hence they are polyclonal with different avidity and affinity. That's why I can't have a reference calibrator with known concentrations and perform a standard curve. 

If so, then I recommend to spend a little more time (and money) to perform titration instead of a single point measurement.

@Antoni I agree that 4PL curve fitting and its inferred IC50 are a good way of comparing antibody titers. However, I run into the same problem: I need a lot of dilutions for single serum sample which I'd like to avoid. Also, if any sera have a very low antibody titer, I might never get a nice 4PL curve as even undiluted samples won't reach saturation, hence impossible to infer IC50. 

I still don't see why a kinetic read on a single dilution won't work if (as explained above in @Edwin), I use the same secondary antibody which recognize the same region on the primary antibody. As pointed out by Rene above, the antibody titer is a combination of concentration and avidity (accumulated strength of multiple affinities) and is method dependent, so even classical titration to LOD is suboptimal and subject to numerous variables. 

Your response suggests that your original question was too vague but the answer is still the same. This will not work. I now understand that you plan to measure the amount of antibody from each sample that binds to an immobilized antigen by measuring the rate of binding of a labeled secondary antibody. My original assumption was that you planned to measure the rate of binding of antibodies in each of the serum samples. Since the relative Ig isotype concentrations in each sample may differ, measuring the rate of binding of a secondary, presumably anti-IgG, antibody, will give misleading results even as to the relative amount of antibody bound to the wells. Sequential dilution of the sera with determination of the amount of antibody bound at equilibrium is required by the very definition of the term, 'titer'.

"Since the relative Ig isotype concentrations in each sample may differ, measuring the rate of binding of a secondary, presumably anti-IgG, antibody, will give misleading results even as to the relative amount of antibody bound to the wells"

Do you mind expanding on this?