Discussion:
Even though there was considerable errors and error bars in the above plot, the credibility of the indicated TTV is strongly suggested by the Lomb-Scargle periodogram (LSP) data (especially with an False-Alarm Probability (FAP) of zero) and the close correspondence of the periodicity of the best-fitting sinusoidal curve to that in the LSP.

TTVs are derived from Q1-Q17 Kepler data.  x-axes: “Observed Tc” (Mid-Transit Time): EXOFAST’s best-fits from Normalized PDCSAP_FLUX Kepler light flux vs. time (BJD_tdb - 2454900) data.  y-axes: “(O-C)”: difference between Observed Tc and the Calculated Tc from the graphically obtained linear ephemeris.

Figure: KOI-823.01, P = 1.03 days [Plot avg. error bars = ± 3.01 min.]
TTV_minimum: 446.65 ± 35.64 days, Amp_ttv_minimum: -0.94 ± 0.14 min.
TTV_maximum: 979.29 ± 49.25 days, Amp_ttv_maximum: 1.18 ± 0.14 min.
TTV_minimum: 1511.92 ± 66.67 days, Amp_ttv_minimum: -0.94 ± 0.14 min.
P_ttv: 1065.28 ± 41.55 days.*
Amp_ttv: 2.12 ± 0.19 minutes.
Lomb-Scargle periodogram, periodicity (P_ttv): 1039.53 days; Power: 41.11; FAP: 0.
Linear ephemeris (this work): Tc = [1.02843546 ± 0.00000013](Tc#) + [103.22954103 ± 0.00009798]


*Szabó et al. reported no such P_ttv for this system using Ford's data for Kepler Q1-5.  Since we were able to use data for Kepler Q1-5, 7-9, 11-13, & 15-17, this is again is another good example of what is discussed on webpage "P_ttvV".

References:
• R. Szabó, Gy. M. Szabó, G. Dálya, A. E. Simon, G. Hodosán, and L. L. Kiss, A & A, 553, A17, May, 2013; arXiv:1207.7229.
• Ford's website: http://www.astro.ufl.edu/~eford/data/kepler/TT_Q0-6/tab_transit_times_electronic.txt
• Numerous additional pertinent literature and major Tc#, Tc, and TTV tabulations can be found on my "Summary" webpage following the table.

6 Feb 2015
Kepler KOI-823 (KIC-5115978) 2-(or more?)-Planet System

Exoplanet-Science.com

Scroll Down  to

KOI's (Blue)  or
K2 Objects (Green)